1"""
   2## Expressions
   3
   4Every AST node in SQLGlot is represented by a subclass of `Expression`.
   5
   6This module contains the implementation of all supported `Expression` types. Additionally,
   7it exposes a number of helper functions, which are mainly used to programmatically build
   8SQL expressions, such as `sqlglot.expressions.select`.
   9
  10----
  11"""
  12
  13from __future__ import annotations
  14
  15import datetime
  16import math
  17import numbers
  18import re
  19import typing as t
  20from collections import deque
  21from copy import deepcopy
  22from enum import auto
  23
  24from sqlglot._typing import E
  25from sqlglot.errors import ParseError
  26from sqlglot.helper import (
  27    AutoName,
  28    camel_to_snake_case,
  29    ensure_collection,
  30    ensure_list,
  31    seq_get,
  32    subclasses,
  33)
  34from sqlglot.tokens import Token
  35
  36if t.TYPE_CHECKING:
  37    from sqlglot.dialects.dialect import DialectType
  38
  39
  40class _Expression(type):
  41    def __new__(cls, clsname, bases, attrs):
  42        klass = super().__new__(cls, clsname, bases, attrs)
  43
  44        # When an Expression class is created, its key is automatically set to be
  45        # the lowercase version of the class' name.
  46        klass.key = clsname.lower()
  47
  48        # This is so that docstrings are not inherited in pdoc
  49        klass.__doc__ = klass.__doc__ or ""
  50
  51        return klass
  52
  53
  54class Expression(metaclass=_Expression):
  55    """
  56    The base class for all expressions in a syntax tree. Each Expression encapsulates any necessary
  57    context, such as its child expressions, their names (arg keys), and whether a given child expression
  58    is optional or not.
  59
  60    Attributes:
  61        key: a unique key for each class in the Expression hierarchy. This is useful for hashing
  62            and representing expressions as strings.
  63        arg_types: determines what arguments (child nodes) are supported by an expression. It
  64            maps arg keys to booleans that indicate whether the corresponding args are optional.
  65        parent: a reference to the parent expression (or None, in case of root expressions).
  66        arg_key: the arg key an expression is associated with, i.e. the name its parent expression
  67            uses to refer to it.
  68        comments: a list of comments that are associated with a given expression. This is used in
  69            order to preserve comments when transpiling SQL code.
  70        _type: the `sqlglot.expressions.DataType` type of an expression. This is inferred by the
  71            optimizer, in order to enable some transformations that require type information.
  72
  73    Example:
  74        >>> class Foo(Expression):
  75        ...     arg_types = {"this": True, "expression": False}
  76
  77        The above definition informs us that Foo is an Expression that requires an argument called
  78        "this" and may also optionally receive an argument called "expression".
  79
  80    Args:
  81        args: a mapping used for retrieving the arguments of an expression, given their arg keys.
  82    """
  83
  84    key = "expression"
  85    arg_types = {"this": True}
  86    __slots__ = ("args", "parent", "arg_key", "comments", "_type", "_meta", "_hash")
  87
  88    def __init__(self, **args: t.Any):
  89        self.args: t.Dict[str, t.Any] = args
  90        self.parent: t.Optional[Expression] = None
  91        self.arg_key: t.Optional[str] = None
  92        self.comments: t.Optional[t.List[str]] = None
  93        self._type: t.Optional[DataType] = None
  94        self._meta: t.Optional[t.Dict[str, t.Any]] = None
  95        self._hash: t.Optional[int] = None
  96
  97        for arg_key, value in self.args.items():
  98            self._set_parent(arg_key, value)
  99
 100    def __eq__(self, other) -> bool:
 101        return type(self) is type(other) and hash(self) == hash(other)
 102
 103    @property
 104    def hashable_args(self) -> t.Any:
 105        args = (self.args.get(k) for k in self.arg_types)
 106
 107        return tuple(
 108            (tuple(_norm_arg(a) for a in arg) if arg else None)
 109            if type(arg) is list
 110            else (_norm_arg(arg) if arg is not None and arg is not False else None)
 111            for arg in args
 112        )
 113
 114    def __hash__(self) -> int:
 115        if self._hash is not None:
 116            return self._hash
 117
 118        return hash((self.__class__, self.hashable_args))
 119
 120    @property
 121    def this(self):
 122        """
 123        Retrieves the argument with key "this".
 124        """
 125        return self.args.get("this")
 126
 127    @property
 128    def expression(self):
 129        """
 130        Retrieves the argument with key "expression".
 131        """
 132        return self.args.get("expression")
 133
 134    @property
 135    def expressions(self):
 136        """
 137        Retrieves the argument with key "expressions".
 138        """
 139        return self.args.get("expressions") or []
 140
 141    def text(self, key) -> str:
 142        """
 143        Returns a textual representation of the argument corresponding to "key". This can only be used
 144        for args that are strings or leaf Expression instances, such as identifiers and literals.
 145        """
 146        field = self.args.get(key)
 147        if isinstance(field, str):
 148            return field
 149        if isinstance(field, (Identifier, Literal, Var)):
 150            return field.this
 151        if isinstance(field, (Star, Null)):
 152            return field.name
 153        return ""
 154
 155    @property
 156    def is_string(self) -> bool:
 157        """
 158        Checks whether a Literal expression is a string.
 159        """
 160        return isinstance(self, Literal) and self.args["is_string"]
 161
 162    @property
 163    def is_number(self) -> bool:
 164        """
 165        Checks whether a Literal expression is a number.
 166        """
 167        return isinstance(self, Literal) and not self.args["is_string"]
 168
 169    @property
 170    def is_int(self) -> bool:
 171        """
 172        Checks whether a Literal expression is an integer.
 173        """
 174        if self.is_number:
 175            try:
 176                int(self.name)
 177                return True
 178            except ValueError:
 179                pass
 180        return False
 181
 182    @property
 183    def is_star(self) -> bool:
 184        """Checks whether an expression is a star."""
 185        return isinstance(self, Star) or (isinstance(self, Column) and isinstance(self.this, Star))
 186
 187    @property
 188    def alias(self) -> str:
 189        """
 190        Returns the alias of the expression, or an empty string if it's not aliased.
 191        """
 192        if isinstance(self.args.get("alias"), TableAlias):
 193            return self.args["alias"].name
 194        return self.text("alias")
 195
 196    @property
 197    def name(self) -> str:
 198        return self.text("this")
 199
 200    @property
 201    def alias_or_name(self) -> str:
 202        return self.alias or self.name
 203
 204    @property
 205    def output_name(self) -> str:
 206        """
 207        Name of the output column if this expression is a selection.
 208
 209        If the Expression has no output name, an empty string is returned.
 210
 211        Example:
 212            >>> from sqlglot import parse_one
 213            >>> parse_one("SELECT a").expressions[0].output_name
 214            'a'
 215            >>> parse_one("SELECT b AS c").expressions[0].output_name
 216            'c'
 217            >>> parse_one("SELECT 1 + 2").expressions[0].output_name
 218            ''
 219        """
 220        return ""
 221
 222    @property
 223    def type(self) -> t.Optional[DataType]:
 224        return self._type
 225
 226    @type.setter
 227    def type(self, dtype: t.Optional[DataType | DataType.Type | str]) -> None:
 228        if dtype and not isinstance(dtype, DataType):
 229            dtype = DataType.build(dtype)
 230        self._type = dtype  # type: ignore
 231
 232    @property
 233    def meta(self) -> t.Dict[str, t.Any]:
 234        if self._meta is None:
 235            self._meta = {}
 236        return self._meta
 237
 238    def __deepcopy__(self, memo):
 239        copy = self.__class__(**deepcopy(self.args))
 240        if self.comments is not None:
 241            copy.comments = deepcopy(self.comments)
 242
 243        if self._type is not None:
 244            copy._type = self._type.copy()
 245
 246        if self._meta is not None:
 247            copy._meta = deepcopy(self._meta)
 248
 249        return copy
 250
 251    def copy(self):
 252        """
 253        Returns a deep copy of the expression.
 254        """
 255        new = deepcopy(self)
 256        new.parent = self.parent
 257        return new
 258
 259    def add_comments(self, comments: t.Optional[t.List[str]]) -> None:
 260        if self.comments is None:
 261            self.comments = []
 262        if comments:
 263            self.comments.extend(comments)
 264
 265    def append(self, arg_key: str, value: t.Any) -> None:
 266        """
 267        Appends value to arg_key if it's a list or sets it as a new list.
 268
 269        Args:
 270            arg_key (str): name of the list expression arg
 271            value (Any): value to append to the list
 272        """
 273        if not isinstance(self.args.get(arg_key), list):
 274            self.args[arg_key] = []
 275        self.args[arg_key].append(value)
 276        self._set_parent(arg_key, value)
 277
 278    def set(self, arg_key: str, value: t.Any) -> None:
 279        """
 280        Sets `arg_key` to `value`.
 281
 282        Args:
 283            arg_key (str): name of the expression arg.
 284            value: value to set the arg to.
 285        """
 286        self.args[arg_key] = value
 287        self._set_parent(arg_key, value)
 288
 289    def _set_parent(self, arg_key: str, value: t.Any) -> None:
 290        if hasattr(value, "parent"):
 291            value.parent = self
 292            value.arg_key = arg_key
 293        elif type(value) is list:
 294            for v in value:
 295                if hasattr(v, "parent"):
 296                    v.parent = self
 297                    v.arg_key = arg_key
 298
 299    @property
 300    def depth(self) -> int:
 301        """
 302        Returns the depth of this tree.
 303        """
 304        if self.parent:
 305            return self.parent.depth + 1
 306        return 0
 307
 308    def iter_expressions(self) -> t.Iterator[t.Tuple[str, Expression]]:
 309        """Yields the key and expression for all arguments, exploding list args."""
 310        for k, vs in self.args.items():
 311            if type(vs) is list:
 312                for v in vs:
 313                    if hasattr(v, "parent"):
 314                        yield k, v
 315            else:
 316                if hasattr(vs, "parent"):
 317                    yield k, vs
 318
 319    def find(self, *expression_types: t.Type[E], bfs: bool = True) -> t.Optional[E]:
 320        """
 321        Returns the first node in this tree which matches at least one of
 322        the specified types.
 323
 324        Args:
 325            expression_types: the expression type(s) to match.
 326            bfs: whether to search the AST using the BFS algorithm (DFS is used if false).
 327
 328        Returns:
 329            The node which matches the criteria or None if no such node was found.
 330        """
 331        return next(self.find_all(*expression_types, bfs=bfs), None)
 332
 333    def find_all(self, *expression_types: t.Type[E], bfs: bool = True) -> t.Iterator[E]:
 334        """
 335        Returns a generator object which visits all nodes in this tree and only
 336        yields those that match at least one of the specified expression types.
 337
 338        Args:
 339            expression_types: the expression type(s) to match.
 340            bfs: whether to search the AST using the BFS algorithm (DFS is used if false).
 341
 342        Returns:
 343            The generator object.
 344        """
 345        for expression, *_ in self.walk(bfs=bfs):
 346            if isinstance(expression, expression_types):
 347                yield expression
 348
 349    def find_ancestor(self, *expression_types: t.Type[E]) -> t.Optional[E]:
 350        """
 351        Returns a nearest parent matching expression_types.
 352
 353        Args:
 354            expression_types: the expression type(s) to match.
 355
 356        Returns:
 357            The parent node.
 358        """
 359        ancestor = self.parent
 360        while ancestor and not isinstance(ancestor, expression_types):
 361            ancestor = ancestor.parent
 362        return t.cast(E, ancestor)
 363
 364    @property
 365    def parent_select(self) -> t.Optional[Select]:
 366        """
 367        Returns the parent select statement.
 368        """
 369        return self.find_ancestor(Select)
 370
 371    @property
 372    def same_parent(self) -> bool:
 373        """Returns if the parent is the same class as itself."""
 374        return type(self.parent) is self.__class__
 375
 376    def root(self) -> Expression:
 377        """
 378        Returns the root expression of this tree.
 379        """
 380        expression = self
 381        while expression.parent:
 382            expression = expression.parent
 383        return expression
 384
 385    def walk(self, bfs=True, prune=None):
 386        """
 387        Returns a generator object which visits all nodes in this tree.
 388
 389        Args:
 390            bfs (bool): if set to True the BFS traversal order will be applied,
 391                otherwise the DFS traversal will be used instead.
 392            prune ((node, parent, arg_key) -> bool): callable that returns True if
 393                the generator should stop traversing this branch of the tree.
 394
 395        Returns:
 396            the generator object.
 397        """
 398        if bfs:
 399            yield from self.bfs(prune=prune)
 400        else:
 401            yield from self.dfs(prune=prune)
 402
 403    def dfs(self, parent=None, key=None, prune=None):
 404        """
 405        Returns a generator object which visits all nodes in this tree in
 406        the DFS (Depth-first) order.
 407
 408        Returns:
 409            The generator object.
 410        """
 411        parent = parent or self.parent
 412        yield self, parent, key
 413        if prune and prune(self, parent, key):
 414            return
 415
 416        for k, v in self.iter_expressions():
 417            yield from v.dfs(self, k, prune)
 418
 419    def bfs(self, prune=None):
 420        """
 421        Returns a generator object which visits all nodes in this tree in
 422        the BFS (Breadth-first) order.
 423
 424        Returns:
 425            The generator object.
 426        """
 427        queue = deque([(self, self.parent, None)])
 428
 429        while queue:
 430            item, parent, key = queue.popleft()
 431
 432            yield item, parent, key
 433            if prune and prune(item, parent, key):
 434                continue
 435
 436            for k, v in item.iter_expressions():
 437                queue.append((v, item, k))
 438
 439    def unnest(self):
 440        """
 441        Returns the first non parenthesis child or self.
 442        """
 443        expression = self
 444        while type(expression) is Paren:
 445            expression = expression.this
 446        return expression
 447
 448    def unalias(self):
 449        """
 450        Returns the inner expression if this is an Alias.
 451        """
 452        if isinstance(self, Alias):
 453            return self.this
 454        return self
 455
 456    def unnest_operands(self):
 457        """
 458        Returns unnested operands as a tuple.
 459        """
 460        return tuple(arg.unnest() for _, arg in self.iter_expressions())
 461
 462    def flatten(self, unnest=True):
 463        """
 464        Returns a generator which yields child nodes who's parents are the same class.
 465
 466        A AND B AND C -> [A, B, C]
 467        """
 468        for node, _, _ in self.dfs(prune=lambda n, p, *_: p and not type(n) is self.__class__):
 469            if not type(node) is self.__class__:
 470                yield node.unnest() if unnest else node
 471
 472    def __str__(self) -> str:
 473        return self.sql()
 474
 475    def __repr__(self) -> str:
 476        return self._to_s()
 477
 478    def sql(self, dialect: DialectType = None, **opts) -> str:
 479        """
 480        Returns SQL string representation of this tree.
 481
 482        Args:
 483            dialect: the dialect of the output SQL string (eg. "spark", "hive", "presto", "mysql").
 484            opts: other `sqlglot.generator.Generator` options.
 485
 486        Returns:
 487            The SQL string.
 488        """
 489        from sqlglot.dialects import Dialect
 490
 491        return Dialect.get_or_raise(dialect)().generate(self, **opts)
 492
 493    def _to_s(self, hide_missing: bool = True, level: int = 0) -> str:
 494        indent = "" if not level else "\n"
 495        indent += "".join(["  "] * level)
 496        left = f"({self.key.upper()} "
 497
 498        args: t.Dict[str, t.Any] = {
 499            k: ", ".join(
 500                v._to_s(hide_missing=hide_missing, level=level + 1)
 501                if hasattr(v, "_to_s")
 502                else str(v)
 503                for v in ensure_list(vs)
 504                if v is not None
 505            )
 506            for k, vs in self.args.items()
 507        }
 508        args["comments"] = self.comments
 509        args["type"] = self.type
 510        args = {k: v for k, v in args.items() if v or not hide_missing}
 511
 512        right = ", ".join(f"{k}: {v}" for k, v in args.items())
 513        right += ")"
 514
 515        return indent + left + right
 516
 517    def transform(self, fun, *args, copy=True, **kwargs):
 518        """
 519        Recursively visits all tree nodes (excluding already transformed ones)
 520        and applies the given transformation function to each node.
 521
 522        Args:
 523            fun (function): a function which takes a node as an argument and returns a
 524                new transformed node or the same node without modifications. If the function
 525                returns None, then the corresponding node will be removed from the syntax tree.
 526            copy (bool): if set to True a new tree instance is constructed, otherwise the tree is
 527                modified in place.
 528
 529        Returns:
 530            The transformed tree.
 531        """
 532        node = self.copy() if copy else self
 533        new_node = fun(node, *args, **kwargs)
 534
 535        if new_node is None or not isinstance(new_node, Expression):
 536            return new_node
 537        if new_node is not node:
 538            new_node.parent = node.parent
 539            return new_node
 540
 541        replace_children(new_node, lambda child: child.transform(fun, *args, copy=False, **kwargs))
 542        return new_node
 543
 544    @t.overload
 545    def replace(self, expression: E) -> E:
 546        ...
 547
 548    @t.overload
 549    def replace(self, expression: None) -> None:
 550        ...
 551
 552    def replace(self, expression):
 553        """
 554        Swap out this expression with a new expression.
 555
 556        For example::
 557
 558            >>> tree = Select().select("x").from_("tbl")
 559            >>> tree.find(Column).replace(Column(this="y"))
 560            (COLUMN this: y)
 561            >>> tree.sql()
 562            'SELECT y FROM tbl'
 563
 564        Args:
 565            expression: new node
 566
 567        Returns:
 568            The new expression or expressions.
 569        """
 570        if not self.parent:
 571            return expression
 572
 573        parent = self.parent
 574        self.parent = None
 575
 576        replace_children(parent, lambda child: expression if child is self else child)
 577        return expression
 578
 579    def pop(self: E) -> E:
 580        """
 581        Remove this expression from its AST.
 582
 583        Returns:
 584            The popped expression.
 585        """
 586        self.replace(None)
 587        return self
 588
 589    def assert_is(self, type_: t.Type[E]) -> E:
 590        """
 591        Assert that this `Expression` is an instance of `type_`.
 592
 593        If it is NOT an instance of `type_`, this raises an assertion error.
 594        Otherwise, this returns this expression.
 595
 596        Examples:
 597            This is useful for type security in chained expressions:
 598
 599            >>> import sqlglot
 600            >>> sqlglot.parse_one("SELECT x from y").assert_is(Select).select("z").sql()
 601            'SELECT x, z FROM y'
 602        """
 603        assert isinstance(self, type_)
 604        return self
 605
 606    def error_messages(self, args: t.Optional[t.Sequence] = None) -> t.List[str]:
 607        """
 608        Checks if this expression is valid (e.g. all mandatory args are set).
 609
 610        Args:
 611            args: a sequence of values that were used to instantiate a Func expression. This is used
 612                to check that the provided arguments don't exceed the function argument limit.
 613
 614        Returns:
 615            A list of error messages for all possible errors that were found.
 616        """
 617        errors: t.List[str] = []
 618
 619        for k in self.args:
 620            if k not in self.arg_types:
 621                errors.append(f"Unexpected keyword: '{k}' for {self.__class__}")
 622        for k, mandatory in self.arg_types.items():
 623            v = self.args.get(k)
 624            if mandatory and (v is None or (isinstance(v, list) and not v)):
 625                errors.append(f"Required keyword: '{k}' missing for {self.__class__}")
 626
 627        if (
 628            args
 629            and isinstance(self, Func)
 630            and len(args) > len(self.arg_types)
 631            and not self.is_var_len_args
 632        ):
 633            errors.append(
 634                f"The number of provided arguments ({len(args)}) is greater than "
 635                f"the maximum number of supported arguments ({len(self.arg_types)})"
 636            )
 637
 638        return errors
 639
 640    def dump(self):
 641        """
 642        Dump this Expression to a JSON-serializable dict.
 643        """
 644        from sqlglot.serde import dump
 645
 646        return dump(self)
 647
 648    @classmethod
 649    def load(cls, obj):
 650        """
 651        Load a dict (as returned by `Expression.dump`) into an Expression instance.
 652        """
 653        from sqlglot.serde import load
 654
 655        return load(obj)
 656
 657
 658IntoType = t.Union[
 659    str,
 660    t.Type[Expression],
 661    t.Collection[t.Union[str, t.Type[Expression]]],
 662]
 663ExpOrStr = t.Union[str, Expression]
 664
 665
 666class Condition(Expression):
 667    def and_(
 668        self,
 669        *expressions: t.Optional[ExpOrStr],
 670        dialect: DialectType = None,
 671        copy: bool = True,
 672        **opts,
 673    ) -> Condition:
 674        """
 675        AND this condition with one or multiple expressions.
 676
 677        Example:
 678            >>> condition("x=1").and_("y=1").sql()
 679            'x = 1 AND y = 1'
 680
 681        Args:
 682            *expressions: the SQL code strings to parse.
 683                If an `Expression` instance is passed, it will be used as-is.
 684            dialect: the dialect used to parse the input expression.
 685            copy: whether or not to copy the involved expressions (only applies to Expressions).
 686            opts: other options to use to parse the input expressions.
 687
 688        Returns:
 689            The new And condition.
 690        """
 691        return and_(self, *expressions, dialect=dialect, copy=copy, **opts)
 692
 693    def or_(
 694        self,
 695        *expressions: t.Optional[ExpOrStr],
 696        dialect: DialectType = None,
 697        copy: bool = True,
 698        **opts,
 699    ) -> Condition:
 700        """
 701        OR this condition with one or multiple expressions.
 702
 703        Example:
 704            >>> condition("x=1").or_("y=1").sql()
 705            'x = 1 OR y = 1'
 706
 707        Args:
 708            *expressions: the SQL code strings to parse.
 709                If an `Expression` instance is passed, it will be used as-is.
 710            dialect: the dialect used to parse the input expression.
 711            copy: whether or not to copy the involved expressions (only applies to Expressions).
 712            opts: other options to use to parse the input expressions.
 713
 714        Returns:
 715            The new Or condition.
 716        """
 717        return or_(self, *expressions, dialect=dialect, copy=copy, **opts)
 718
 719    def not_(self, copy: bool = True):
 720        """
 721        Wrap this condition with NOT.
 722
 723        Example:
 724            >>> condition("x=1").not_().sql()
 725            'NOT x = 1'
 726
 727        Args:
 728            copy: whether or not to copy this object.
 729
 730        Returns:
 731            The new Not instance.
 732        """
 733        return not_(self, copy=copy)
 734
 735    def as_(
 736        self,
 737        alias: str | Identifier,
 738        quoted: t.Optional[bool] = None,
 739        dialect: DialectType = None,
 740        copy: bool = True,
 741        **opts,
 742    ) -> Alias:
 743        return alias_(self, alias, quoted=quoted, dialect=dialect, copy=copy, **opts)
 744
 745    def _binop(self, klass: t.Type[E], other: t.Any, reverse: bool = False) -> E:
 746        this = self.copy()
 747        other = convert(other, copy=True)
 748        if not isinstance(this, klass) and not isinstance(other, klass):
 749            this = _wrap(this, Binary)
 750            other = _wrap(other, Binary)
 751        if reverse:
 752            return klass(this=other, expression=this)
 753        return klass(this=this, expression=other)
 754
 755    def __getitem__(self, other: ExpOrStr | t.Tuple[ExpOrStr]):
 756        return Bracket(
 757            this=self.copy(), expressions=[convert(e, copy=True) for e in ensure_list(other)]
 758        )
 759
 760    def isin(
 761        self, *expressions: t.Any, query: t.Optional[ExpOrStr] = None, copy: bool = True, **opts
 762    ) -> In:
 763        return In(
 764            this=_maybe_copy(self, copy),
 765            expressions=[convert(e, copy=copy) for e in expressions],
 766            query=maybe_parse(query, copy=copy, **opts) if query else None,
 767        )
 768
 769    def between(self, low: t.Any, high: t.Any, copy: bool = True, **opts) -> Between:
 770        return Between(
 771            this=_maybe_copy(self, copy),
 772            low=convert(low, copy=copy, **opts),
 773            high=convert(high, copy=copy, **opts),
 774        )
 775
 776    def is_(self, other: ExpOrStr) -> Is:
 777        return self._binop(Is, other)
 778
 779    def like(self, other: ExpOrStr) -> Like:
 780        return self._binop(Like, other)
 781
 782    def ilike(self, other: ExpOrStr) -> ILike:
 783        return self._binop(ILike, other)
 784
 785    def eq(self, other: t.Any) -> EQ:
 786        return self._binop(EQ, other)
 787
 788    def neq(self, other: t.Any) -> NEQ:
 789        return self._binop(NEQ, other)
 790
 791    def rlike(self, other: ExpOrStr) -> RegexpLike:
 792        return self._binop(RegexpLike, other)
 793
 794    def __lt__(self, other: t.Any) -> LT:
 795        return self._binop(LT, other)
 796
 797    def __le__(self, other: t.Any) -> LTE:
 798        return self._binop(LTE, other)
 799
 800    def __gt__(self, other: t.Any) -> GT:
 801        return self._binop(GT, other)
 802
 803    def __ge__(self, other: t.Any) -> GTE:
 804        return self._binop(GTE, other)
 805
 806    def __add__(self, other: t.Any) -> Add:
 807        return self._binop(Add, other)
 808
 809    def __radd__(self, other: t.Any) -> Add:
 810        return self._binop(Add, other, reverse=True)
 811
 812    def __sub__(self, other: t.Any) -> Sub:
 813        return self._binop(Sub, other)
 814
 815    def __rsub__(self, other: t.Any) -> Sub:
 816        return self._binop(Sub, other, reverse=True)
 817
 818    def __mul__(self, other: t.Any) -> Mul:
 819        return self._binop(Mul, other)
 820
 821    def __rmul__(self, other: t.Any) -> Mul:
 822        return self._binop(Mul, other, reverse=True)
 823
 824    def __truediv__(self, other: t.Any) -> Div:
 825        return self._binop(Div, other)
 826
 827    def __rtruediv__(self, other: t.Any) -> Div:
 828        return self._binop(Div, other, reverse=True)
 829
 830    def __floordiv__(self, other: t.Any) -> IntDiv:
 831        return self._binop(IntDiv, other)
 832
 833    def __rfloordiv__(self, other: t.Any) -> IntDiv:
 834        return self._binop(IntDiv, other, reverse=True)
 835
 836    def __mod__(self, other: t.Any) -> Mod:
 837        return self._binop(Mod, other)
 838
 839    def __rmod__(self, other: t.Any) -> Mod:
 840        return self._binop(Mod, other, reverse=True)
 841
 842    def __pow__(self, other: t.Any) -> Pow:
 843        return self._binop(Pow, other)
 844
 845    def __rpow__(self, other: t.Any) -> Pow:
 846        return self._binop(Pow, other, reverse=True)
 847
 848    def __and__(self, other: t.Any) -> And:
 849        return self._binop(And, other)
 850
 851    def __rand__(self, other: t.Any) -> And:
 852        return self._binop(And, other, reverse=True)
 853
 854    def __or__(self, other: t.Any) -> Or:
 855        return self._binop(Or, other)
 856
 857    def __ror__(self, other: t.Any) -> Or:
 858        return self._binop(Or, other, reverse=True)
 859
 860    def __neg__(self) -> Neg:
 861        return Neg(this=_wrap(self.copy(), Binary))
 862
 863    def __invert__(self) -> Not:
 864        return not_(self.copy())
 865
 866
 867class Predicate(Condition):
 868    """Relationships like x = y, x > 1, x >= y."""
 869
 870
 871class DerivedTable(Expression):
 872    @property
 873    def alias_column_names(self) -> t.List[str]:
 874        table_alias = self.args.get("alias")
 875        if not table_alias:
 876            return []
 877        return [c.name for c in table_alias.args.get("columns") or []]
 878
 879    @property
 880    def selects(self):
 881        return self.this.selects if isinstance(self.this, Subqueryable) else []
 882
 883    @property
 884    def named_selects(self):
 885        return [select.output_name for select in self.selects]
 886
 887
 888class Unionable(Expression):
 889    def union(
 890        self, expression: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
 891    ) -> Unionable:
 892        """
 893        Builds a UNION expression.
 894
 895        Example:
 896            >>> import sqlglot
 897            >>> sqlglot.parse_one("SELECT * FROM foo").union("SELECT * FROM bla").sql()
 898            'SELECT * FROM foo UNION SELECT * FROM bla'
 899
 900        Args:
 901            expression: the SQL code string.
 902                If an `Expression` instance is passed, it will be used as-is.
 903            distinct: set the DISTINCT flag if and only if this is true.
 904            dialect: the dialect used to parse the input expression.
 905            opts: other options to use to parse the input expressions.
 906
 907        Returns:
 908            The new Union expression.
 909        """
 910        return union(left=self, right=expression, distinct=distinct, dialect=dialect, **opts)
 911
 912    def intersect(
 913        self, expression: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
 914    ) -> Unionable:
 915        """
 916        Builds an INTERSECT expression.
 917
 918        Example:
 919            >>> import sqlglot
 920            >>> sqlglot.parse_one("SELECT * FROM foo").intersect("SELECT * FROM bla").sql()
 921            'SELECT * FROM foo INTERSECT SELECT * FROM bla'
 922
 923        Args:
 924            expression: the SQL code string.
 925                If an `Expression` instance is passed, it will be used as-is.
 926            distinct: set the DISTINCT flag if and only if this is true.
 927            dialect: the dialect used to parse the input expression.
 928            opts: other options to use to parse the input expressions.
 929
 930        Returns:
 931            The new Intersect expression.
 932        """
 933        return intersect(left=self, right=expression, distinct=distinct, dialect=dialect, **opts)
 934
 935    def except_(
 936        self, expression: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
 937    ) -> Unionable:
 938        """
 939        Builds an EXCEPT expression.
 940
 941        Example:
 942            >>> import sqlglot
 943            >>> sqlglot.parse_one("SELECT * FROM foo").except_("SELECT * FROM bla").sql()
 944            'SELECT * FROM foo EXCEPT SELECT * FROM bla'
 945
 946        Args:
 947            expression: the SQL code string.
 948                If an `Expression` instance is passed, it will be used as-is.
 949            distinct: set the DISTINCT flag if and only if this is true.
 950            dialect: the dialect used to parse the input expression.
 951            opts: other options to use to parse the input expressions.
 952
 953        Returns:
 954            The new Except expression.
 955        """
 956        return except_(left=self, right=expression, distinct=distinct, dialect=dialect, **opts)
 957
 958
 959class UDTF(DerivedTable, Unionable):
 960    @property
 961    def selects(self):
 962        alias = self.args.get("alias")
 963        return alias.columns if alias else []
 964
 965
 966class Cache(Expression):
 967    arg_types = {
 968        "with": False,
 969        "this": True,
 970        "lazy": False,
 971        "options": False,
 972        "expression": False,
 973    }
 974
 975
 976class Uncache(Expression):
 977    arg_types = {"this": True, "exists": False}
 978
 979
 980class Create(Expression):
 981    arg_types = {
 982        "with": False,
 983        "this": True,
 984        "kind": True,
 985        "expression": False,
 986        "exists": False,
 987        "properties": False,
 988        "replace": False,
 989        "unique": False,
 990        "indexes": False,
 991        "no_schema_binding": False,
 992        "begin": False,
 993        "clone": False,
 994    }
 995
 996
 997# https://docs.snowflake.com/en/sql-reference/sql/create-clone
 998class Clone(Expression):
 999    arg_types = {
1000        "this": True,
1001        "when": False,
1002        "kind": False,
1003        "expression": False,
1004    }
1005
1006
1007class Describe(Expression):
1008    arg_types = {"this": True, "kind": False}
1009
1010
1011class Pragma(Expression):
1012    pass
1013
1014
1015class Set(Expression):
1016    arg_types = {"expressions": False, "unset": False, "tag": False}
1017
1018
1019class SetItem(Expression):
1020    arg_types = {
1021        "this": False,
1022        "expressions": False,
1023        "kind": False,
1024        "collate": False,  # MySQL SET NAMES statement
1025        "global": False,
1026    }
1027
1028
1029class Show(Expression):
1030    arg_types = {
1031        "this": True,
1032        "target": False,
1033        "offset": False,
1034        "limit": False,
1035        "like": False,
1036        "where": False,
1037        "db": False,
1038        "full": False,
1039        "mutex": False,
1040        "query": False,
1041        "channel": False,
1042        "global": False,
1043        "log": False,
1044        "position": False,
1045        "types": False,
1046    }
1047
1048
1049class UserDefinedFunction(Expression):
1050    arg_types = {"this": True, "expressions": False, "wrapped": False}
1051
1052
1053class CharacterSet(Expression):
1054    arg_types = {"this": True, "default": False}
1055
1056
1057class With(Expression):
1058    arg_types = {"expressions": True, "recursive": False}
1059
1060    @property
1061    def recursive(self) -> bool:
1062        return bool(self.args.get("recursive"))
1063
1064
1065class WithinGroup(Expression):
1066    arg_types = {"this": True, "expression": False}
1067
1068
1069class CTE(DerivedTable):
1070    arg_types = {"this": True, "alias": True}
1071
1072
1073class TableAlias(Expression):
1074    arg_types = {"this": False, "columns": False}
1075
1076    @property
1077    def columns(self):
1078        return self.args.get("columns") or []
1079
1080
1081class BitString(Condition):
1082    pass
1083
1084
1085class HexString(Condition):
1086    pass
1087
1088
1089class ByteString(Condition):
1090    pass
1091
1092
1093class RawString(Condition):
1094    pass
1095
1096
1097class Column(Condition):
1098    arg_types = {"this": True, "table": False, "db": False, "catalog": False, "join_mark": False}
1099
1100    @property
1101    def table(self) -> str:
1102        return self.text("table")
1103
1104    @property
1105    def db(self) -> str:
1106        return self.text("db")
1107
1108    @property
1109    def catalog(self) -> str:
1110        return self.text("catalog")
1111
1112    @property
1113    def output_name(self) -> str:
1114        return self.name
1115
1116    @property
1117    def parts(self) -> t.List[Identifier]:
1118        """Return the parts of a column in order catalog, db, table, name."""
1119        return [
1120            t.cast(Identifier, self.args[part])
1121            for part in ("catalog", "db", "table", "this")
1122            if self.args.get(part)
1123        ]
1124
1125    def to_dot(self) -> Dot:
1126        """Converts the column into a dot expression."""
1127        parts = self.parts
1128        parent = self.parent
1129
1130        while parent:
1131            if isinstance(parent, Dot):
1132                parts.append(parent.expression)
1133            parent = parent.parent
1134
1135        return Dot.build(parts)
1136
1137
1138class ColumnPosition(Expression):
1139    arg_types = {"this": False, "position": True}
1140
1141
1142class ColumnDef(Expression):
1143    arg_types = {
1144        "this": True,
1145        "kind": False,
1146        "constraints": False,
1147        "exists": False,
1148        "position": False,
1149    }
1150
1151    @property
1152    def constraints(self) -> t.List[ColumnConstraint]:
1153        return self.args.get("constraints") or []
1154
1155
1156class AlterColumn(Expression):
1157    arg_types = {
1158        "this": True,
1159        "dtype": False,
1160        "collate": False,
1161        "using": False,
1162        "default": False,
1163        "drop": False,
1164    }
1165
1166
1167class RenameTable(Expression):
1168    pass
1169
1170
1171class Comment(Expression):
1172    arg_types = {"this": True, "kind": True, "expression": True, "exists": False}
1173
1174
1175# https://clickhouse.com/docs/en/engines/table-engines/mergetree-family/mergetree#mergetree-table-ttl
1176class MergeTreeTTLAction(Expression):
1177    arg_types = {
1178        "this": True,
1179        "delete": False,
1180        "recompress": False,
1181        "to_disk": False,
1182        "to_volume": False,
1183    }
1184
1185
1186# https://clickhouse.com/docs/en/engines/table-engines/mergetree-family/mergetree#mergetree-table-ttl
1187class MergeTreeTTL(Expression):
1188    arg_types = {
1189        "expressions": True,
1190        "where": False,
1191        "group": False,
1192        "aggregates": False,
1193    }
1194
1195
1196class ColumnConstraint(Expression):
1197    arg_types = {"this": False, "kind": True}
1198
1199    @property
1200    def kind(self) -> ColumnConstraintKind:
1201        return self.args["kind"]
1202
1203
1204class ColumnConstraintKind(Expression):
1205    pass
1206
1207
1208class AutoIncrementColumnConstraint(ColumnConstraintKind):
1209    pass
1210
1211
1212class CaseSpecificColumnConstraint(ColumnConstraintKind):
1213    arg_types = {"not_": True}
1214
1215
1216class CharacterSetColumnConstraint(ColumnConstraintKind):
1217    arg_types = {"this": True}
1218
1219
1220class CheckColumnConstraint(ColumnConstraintKind):
1221    pass
1222
1223
1224class CollateColumnConstraint(ColumnConstraintKind):
1225    pass
1226
1227
1228class CommentColumnConstraint(ColumnConstraintKind):
1229    pass
1230
1231
1232class CompressColumnConstraint(ColumnConstraintKind):
1233    pass
1234
1235
1236class DateFormatColumnConstraint(ColumnConstraintKind):
1237    arg_types = {"this": True}
1238
1239
1240class DefaultColumnConstraint(ColumnConstraintKind):
1241    pass
1242
1243
1244class EncodeColumnConstraint(ColumnConstraintKind):
1245    pass
1246
1247
1248class GeneratedAsIdentityColumnConstraint(ColumnConstraintKind):
1249    # this: True -> ALWAYS, this: False -> BY DEFAULT
1250    arg_types = {
1251        "this": False,
1252        "expression": False,
1253        "on_null": False,
1254        "start": False,
1255        "increment": False,
1256        "minvalue": False,
1257        "maxvalue": False,
1258        "cycle": False,
1259    }
1260
1261
1262class InlineLengthColumnConstraint(ColumnConstraintKind):
1263    pass
1264
1265
1266class NotNullColumnConstraint(ColumnConstraintKind):
1267    arg_types = {"allow_null": False}
1268
1269
1270# https://dev.mysql.com/doc/refman/5.7/en/timestamp-initialization.html
1271class OnUpdateColumnConstraint(ColumnConstraintKind):
1272    pass
1273
1274
1275class PrimaryKeyColumnConstraint(ColumnConstraintKind):
1276    arg_types = {"desc": False}
1277
1278
1279class TitleColumnConstraint(ColumnConstraintKind):
1280    pass
1281
1282
1283class UniqueColumnConstraint(ColumnConstraintKind):
1284    arg_types = {"this": False}
1285
1286
1287class UppercaseColumnConstraint(ColumnConstraintKind):
1288    arg_types: t.Dict[str, t.Any] = {}
1289
1290
1291class PathColumnConstraint(ColumnConstraintKind):
1292    pass
1293
1294
1295class Constraint(Expression):
1296    arg_types = {"this": True, "expressions": True}
1297
1298
1299class Delete(Expression):
1300    arg_types = {
1301        "with": False,
1302        "this": False,
1303        "using": False,
1304        "where": False,
1305        "returning": False,
1306        "limit": False,
1307    }
1308
1309    def delete(
1310        self,
1311        table: ExpOrStr,
1312        dialect: DialectType = None,
1313        copy: bool = True,
1314        **opts,
1315    ) -> Delete:
1316        """
1317        Create a DELETE expression or replace the table on an existing DELETE expression.
1318
1319        Example:
1320            >>> delete("tbl").sql()
1321            'DELETE FROM tbl'
1322
1323        Args:
1324            table: the table from which to delete.
1325            dialect: the dialect used to parse the input expression.
1326            copy: if `False`, modify this expression instance in-place.
1327            opts: other options to use to parse the input expressions.
1328
1329        Returns:
1330            Delete: the modified expression.
1331        """
1332        return _apply_builder(
1333            expression=table,
1334            instance=self,
1335            arg="this",
1336            dialect=dialect,
1337            into=Table,
1338            copy=copy,
1339            **opts,
1340        )
1341
1342    def where(
1343        self,
1344        *expressions: t.Optional[ExpOrStr],
1345        append: bool = True,
1346        dialect: DialectType = None,
1347        copy: bool = True,
1348        **opts,
1349    ) -> Delete:
1350        """
1351        Append to or set the WHERE expressions.
1352
1353        Example:
1354            >>> delete("tbl").where("x = 'a' OR x < 'b'").sql()
1355            "DELETE FROM tbl WHERE x = 'a' OR x < 'b'"
1356
1357        Args:
1358            *expressions: the SQL code strings to parse.
1359                If an `Expression` instance is passed, it will be used as-is.
1360                Multiple expressions are combined with an AND operator.
1361            append: if `True`, AND the new expressions to any existing expression.
1362                Otherwise, this resets the expression.
1363            dialect: the dialect used to parse the input expressions.
1364            copy: if `False`, modify this expression instance in-place.
1365            opts: other options to use to parse the input expressions.
1366
1367        Returns:
1368            Delete: the modified expression.
1369        """
1370        return _apply_conjunction_builder(
1371            *expressions,
1372            instance=self,
1373            arg="where",
1374            append=append,
1375            into=Where,
1376            dialect=dialect,
1377            copy=copy,
1378            **opts,
1379        )
1380
1381    def returning(
1382        self,
1383        expression: ExpOrStr,
1384        dialect: DialectType = None,
1385        copy: bool = True,
1386        **opts,
1387    ) -> Delete:
1388        """
1389        Set the RETURNING expression. Not supported by all dialects.
1390
1391        Example:
1392            >>> delete("tbl").returning("*", dialect="postgres").sql()
1393            'DELETE FROM tbl RETURNING *'
1394
1395        Args:
1396            expression: the SQL code strings to parse.
1397                If an `Expression` instance is passed, it will be used as-is.
1398            dialect: the dialect used to parse the input expressions.
1399            copy: if `False`, modify this expression instance in-place.
1400            opts: other options to use to parse the input expressions.
1401
1402        Returns:
1403            Delete: the modified expression.
1404        """
1405        return _apply_builder(
1406            expression=expression,
1407            instance=self,
1408            arg="returning",
1409            prefix="RETURNING",
1410            dialect=dialect,
1411            copy=copy,
1412            into=Returning,
1413            **opts,
1414        )
1415
1416
1417class Drop(Expression):
1418    arg_types = {
1419        "this": False,
1420        "kind": False,
1421        "exists": False,
1422        "temporary": False,
1423        "materialized": False,
1424        "cascade": False,
1425        "constraints": False,
1426        "purge": False,
1427    }
1428
1429
1430class Filter(Expression):
1431    arg_types = {"this": True, "expression": True}
1432
1433
1434class Check(Expression):
1435    pass
1436
1437
1438class Directory(Expression):
1439    # https://spark.apache.org/docs/3.0.0-preview/sql-ref-syntax-dml-insert-overwrite-directory-hive.html
1440    arg_types = {"this": True, "local": False, "row_format": False}
1441
1442
1443class ForeignKey(Expression):
1444    arg_types = {
1445        "expressions": True,
1446        "reference": False,
1447        "delete": False,
1448        "update": False,
1449    }
1450
1451
1452class PrimaryKey(Expression):
1453    arg_types = {"expressions": True, "options": False}
1454
1455
1456# https://www.postgresql.org/docs/9.1/sql-selectinto.html
1457# https://docs.aws.amazon.com/redshift/latest/dg/r_SELECT_INTO.html#r_SELECT_INTO-examples
1458class Into(Expression):
1459    arg_types = {"this": True, "temporary": False, "unlogged": False}
1460
1461
1462class From(Expression):
1463    @property
1464    def name(self) -> str:
1465        return self.this.name
1466
1467    @property
1468    def alias_or_name(self) -> str:
1469        return self.this.alias_or_name
1470
1471
1472class Having(Expression):
1473    pass
1474
1475
1476class Hint(Expression):
1477    arg_types = {"expressions": True}
1478
1479
1480class JoinHint(Expression):
1481    arg_types = {"this": True, "expressions": True}
1482
1483
1484class Identifier(Expression):
1485    arg_types = {"this": True, "quoted": False}
1486
1487    @property
1488    def quoted(self) -> bool:
1489        return bool(self.args.get("quoted"))
1490
1491    @property
1492    def hashable_args(self) -> t.Any:
1493        if self.quoted and any(char.isupper() for char in self.this):
1494            return (self.this, self.quoted)
1495        return self.this.lower()
1496
1497    @property
1498    def output_name(self) -> str:
1499        return self.name
1500
1501
1502class Index(Expression):
1503    arg_types = {
1504        "this": False,
1505        "table": False,
1506        "using": False,
1507        "where": False,
1508        "columns": False,
1509        "unique": False,
1510        "primary": False,
1511        "amp": False,  # teradata
1512        "partition_by": False,  # teradata
1513    }
1514
1515
1516class Insert(Expression):
1517    arg_types = {
1518        "with": False,
1519        "this": True,
1520        "expression": False,
1521        "conflict": False,
1522        "returning": False,
1523        "overwrite": False,
1524        "exists": False,
1525        "partition": False,
1526        "alternative": False,
1527        "where": False,
1528    }
1529
1530    def with_(
1531        self,
1532        alias: ExpOrStr,
1533        as_: ExpOrStr,
1534        recursive: t.Optional[bool] = None,
1535        append: bool = True,
1536        dialect: DialectType = None,
1537        copy: bool = True,
1538        **opts,
1539    ) -> Insert:
1540        """
1541        Append to or set the common table expressions.
1542
1543        Example:
1544            >>> insert("SELECT x FROM cte", "t").with_("cte", as_="SELECT * FROM tbl").sql()
1545            'WITH cte AS (SELECT * FROM tbl) INSERT INTO t SELECT x FROM cte'
1546
1547        Args:
1548            alias: the SQL code string to parse as the table name.
1549                If an `Expression` instance is passed, this is used as-is.
1550            as_: the SQL code string to parse as the table expression.
1551                If an `Expression` instance is passed, it will be used as-is.
1552            recursive: set the RECURSIVE part of the expression. Defaults to `False`.
1553            append: if `True`, add to any existing expressions.
1554                Otherwise, this resets the expressions.
1555            dialect: the dialect used to parse the input expression.
1556            copy: if `False`, modify this expression instance in-place.
1557            opts: other options to use to parse the input expressions.
1558
1559        Returns:
1560            The modified expression.
1561        """
1562        return _apply_cte_builder(
1563            self, alias, as_, recursive=recursive, append=append, dialect=dialect, copy=copy, **opts
1564        )
1565
1566
1567class OnConflict(Expression):
1568    arg_types = {
1569        "duplicate": False,
1570        "expressions": False,
1571        "nothing": False,
1572        "key": False,
1573        "constraint": False,
1574    }
1575
1576
1577class Returning(Expression):
1578    arg_types = {"expressions": True}
1579
1580
1581# https://dev.mysql.com/doc/refman/8.0/en/charset-introducer.html
1582class Introducer(Expression):
1583    arg_types = {"this": True, "expression": True}
1584
1585
1586# national char, like n'utf8'
1587class National(Expression):
1588    pass
1589
1590
1591class LoadData(Expression):
1592    arg_types = {
1593        "this": True,
1594        "local": False,
1595        "overwrite": False,
1596        "inpath": True,
1597        "partition": False,
1598        "input_format": False,
1599        "serde": False,
1600    }
1601
1602
1603class Partition(Expression):
1604    arg_types = {"expressions": True}
1605
1606
1607class Fetch(Expression):
1608    arg_types = {
1609        "direction": False,
1610        "count": False,
1611        "percent": False,
1612        "with_ties": False,
1613    }
1614
1615
1616class Group(Expression):
1617    arg_types = {
1618        "expressions": False,
1619        "grouping_sets": False,
1620        "cube": False,
1621        "rollup": False,
1622        "totals": False,
1623    }
1624
1625
1626class Lambda(Expression):
1627    arg_types = {"this": True, "expressions": True}
1628
1629
1630class Limit(Expression):
1631    arg_types = {"this": False, "expression": True, "offset": False}
1632
1633
1634class Literal(Condition):
1635    arg_types = {"this": True, "is_string": True}
1636
1637    @property
1638    def hashable_args(self) -> t.Any:
1639        return (self.this, self.args.get("is_string"))
1640
1641    @classmethod
1642    def number(cls, number) -> Literal:
1643        return cls(this=str(number), is_string=False)
1644
1645    @classmethod
1646    def string(cls, string) -> Literal:
1647        return cls(this=str(string), is_string=True)
1648
1649    @property
1650    def output_name(self) -> str:
1651        return self.name
1652
1653
1654class Join(Expression):
1655    arg_types = {
1656        "this": True,
1657        "on": False,
1658        "side": False,
1659        "kind": False,
1660        "using": False,
1661        "method": False,
1662        "global": False,
1663        "hint": False,
1664    }
1665
1666    @property
1667    def method(self) -> str:
1668        return self.text("method").upper()
1669
1670    @property
1671    def kind(self) -> str:
1672        return self.text("kind").upper()
1673
1674    @property
1675    def side(self) -> str:
1676        return self.text("side").upper()
1677
1678    @property
1679    def hint(self) -> str:
1680        return self.text("hint").upper()
1681
1682    @property
1683    def alias_or_name(self) -> str:
1684        return self.this.alias_or_name
1685
1686    def on(
1687        self,
1688        *expressions: t.Optional[ExpOrStr],
1689        append: bool = True,
1690        dialect: DialectType = None,
1691        copy: bool = True,
1692        **opts,
1693    ) -> Join:
1694        """
1695        Append to or set the ON expressions.
1696
1697        Example:
1698            >>> import sqlglot
1699            >>> sqlglot.parse_one("JOIN x", into=Join).on("y = 1").sql()
1700            'JOIN x ON y = 1'
1701
1702        Args:
1703            *expressions: the SQL code strings to parse.
1704                If an `Expression` instance is passed, it will be used as-is.
1705                Multiple expressions are combined with an AND operator.
1706            append: if `True`, AND the new expressions to any existing expression.
1707                Otherwise, this resets the expression.
1708            dialect: the dialect used to parse the input expressions.
1709            copy: if `False`, modify this expression instance in-place.
1710            opts: other options to use to parse the input expressions.
1711
1712        Returns:
1713            The modified Join expression.
1714        """
1715        join = _apply_conjunction_builder(
1716            *expressions,
1717            instance=self,
1718            arg="on",
1719            append=append,
1720            dialect=dialect,
1721            copy=copy,
1722            **opts,
1723        )
1724
1725        if join.kind == "CROSS":
1726            join.set("kind", None)
1727
1728        return join
1729
1730    def using(
1731        self,
1732        *expressions: t.Optional[ExpOrStr],
1733        append: bool = True,
1734        dialect: DialectType = None,
1735        copy: bool = True,
1736        **opts,
1737    ) -> Join:
1738        """
1739        Append to or set the USING expressions.
1740
1741        Example:
1742            >>> import sqlglot
1743            >>> sqlglot.parse_one("JOIN x", into=Join).using("foo", "bla").sql()
1744            'JOIN x USING (foo, bla)'
1745
1746        Args:
1747            *expressions: the SQL code strings to parse.
1748                If an `Expression` instance is passed, it will be used as-is.
1749            append: if `True`, concatenate the new expressions to the existing "using" list.
1750                Otherwise, this resets the expression.
1751            dialect: the dialect used to parse the input expressions.
1752            copy: if `False`, modify this expression instance in-place.
1753            opts: other options to use to parse the input expressions.
1754
1755        Returns:
1756            The modified Join expression.
1757        """
1758        join = _apply_list_builder(
1759            *expressions,
1760            instance=self,
1761            arg="using",
1762            append=append,
1763            dialect=dialect,
1764            copy=copy,
1765            **opts,
1766        )
1767
1768        if join.kind == "CROSS":
1769            join.set("kind", None)
1770
1771        return join
1772
1773
1774class Lateral(UDTF):
1775    arg_types = {"this": True, "view": False, "outer": False, "alias": False}
1776
1777
1778class MatchRecognize(Expression):
1779    arg_types = {
1780        "partition_by": False,
1781        "order": False,
1782        "measures": False,
1783        "rows": False,
1784        "after": False,
1785        "pattern": False,
1786        "define": False,
1787        "alias": False,
1788    }
1789
1790
1791# Clickhouse FROM FINAL modifier
1792# https://clickhouse.com/docs/en/sql-reference/statements/select/from/#final-modifier
1793class Final(Expression):
1794    pass
1795
1796
1797class Offset(Expression):
1798    arg_types = {"this": False, "expression": True}
1799
1800
1801class Order(Expression):
1802    arg_types = {"this": False, "expressions": True}
1803
1804
1805# hive specific sorts
1806# https://cwiki.apache.org/confluence/display/Hive/LanguageManual+SortBy
1807class Cluster(Order):
1808    pass
1809
1810
1811class Distribute(Order):
1812    pass
1813
1814
1815class Sort(Order):
1816    pass
1817
1818
1819class Ordered(Expression):
1820    arg_types = {"this": True, "desc": True, "nulls_first": True}
1821
1822
1823class Property(Expression):
1824    arg_types = {"this": True, "value": True}
1825
1826
1827class AlgorithmProperty(Property):
1828    arg_types = {"this": True}
1829
1830
1831class AutoIncrementProperty(Property):
1832    arg_types = {"this": True}
1833
1834
1835class BlockCompressionProperty(Property):
1836    arg_types = {"autotemp": False, "always": False, "default": True, "manual": True, "never": True}
1837
1838
1839class CharacterSetProperty(Property):
1840    arg_types = {"this": True, "default": True}
1841
1842
1843class ChecksumProperty(Property):
1844    arg_types = {"on": False, "default": False}
1845
1846
1847class CollateProperty(Property):
1848    arg_types = {"this": True}
1849
1850
1851class CopyGrantsProperty(Property):
1852    arg_types = {}
1853
1854
1855class DataBlocksizeProperty(Property):
1856    arg_types = {
1857        "size": False,
1858        "units": False,
1859        "minimum": False,
1860        "maximum": False,
1861        "default": False,
1862    }
1863
1864
1865class DefinerProperty(Property):
1866    arg_types = {"this": True}
1867
1868
1869class DistKeyProperty(Property):
1870    arg_types = {"this": True}
1871
1872
1873class DistStyleProperty(Property):
1874    arg_types = {"this": True}
1875
1876
1877class EngineProperty(Property):
1878    arg_types = {"this": True}
1879
1880
1881class ToTableProperty(Property):
1882    arg_types = {"this": True}
1883
1884
1885class ExecuteAsProperty(Property):
1886    arg_types = {"this": True}
1887
1888
1889class ExternalProperty(Property):
1890    arg_types = {"this": False}
1891
1892
1893class FallbackProperty(Property):
1894    arg_types = {"no": True, "protection": False}
1895
1896
1897class FileFormatProperty(Property):
1898    arg_types = {"this": True}
1899
1900
1901class FreespaceProperty(Property):
1902    arg_types = {"this": True, "percent": False}
1903
1904
1905class InputOutputFormat(Expression):
1906    arg_types = {"input_format": False, "output_format": False}
1907
1908
1909class IsolatedLoadingProperty(Property):
1910    arg_types = {
1911        "no": True,
1912        "concurrent": True,
1913        "for_all": True,
1914        "for_insert": True,
1915        "for_none": True,
1916    }
1917
1918
1919class JournalProperty(Property):
1920    arg_types = {
1921        "no": False,
1922        "dual": False,
1923        "before": False,
1924        "local": False,
1925        "after": False,
1926    }
1927
1928
1929class LanguageProperty(Property):
1930    arg_types = {"this": True}
1931
1932
1933# spark ddl
1934class ClusteredByProperty(Property):
1935    arg_types = {"expressions": True, "sorted_by": False, "buckets": True}
1936
1937
1938class DictProperty(Property):
1939    arg_types = {"this": True, "kind": True, "settings": False}
1940
1941
1942class DictSubProperty(Property):
1943    pass
1944
1945
1946class DictRange(Property):
1947    arg_types = {"this": True, "min": True, "max": True}
1948
1949
1950# Clickhouse CREATE ... ON CLUSTER modifier
1951# https://clickhouse.com/docs/en/sql-reference/distributed-ddl
1952class OnCluster(Property):
1953    arg_types = {"this": True}
1954
1955
1956class LikeProperty(Property):
1957    arg_types = {"this": True, "expressions": False}
1958
1959
1960class LocationProperty(Property):
1961    arg_types = {"this": True}
1962
1963
1964class LockingProperty(Property):
1965    arg_types = {
1966        "this": False,
1967        "kind": True,
1968        "for_or_in": True,
1969        "lock_type": True,
1970        "override": False,
1971    }
1972
1973
1974class LogProperty(Property):
1975    arg_types = {"no": True}
1976
1977
1978class MaterializedProperty(Property):
1979    arg_types = {"this": False}
1980
1981
1982class MergeBlockRatioProperty(Property):
1983    arg_types = {"this": False, "no": False, "default": False, "percent": False}
1984
1985
1986class NoPrimaryIndexProperty(Property):
1987    arg_types = {}
1988
1989
1990class OnCommitProperty(Property):
1991    arg_type = {"delete": False}
1992
1993
1994class PartitionedByProperty(Property):
1995    arg_types = {"this": True}
1996
1997
1998class ReturnsProperty(Property):
1999    arg_types = {"this": True, "is_table": False, "table": False}
2000
2001
2002class RowFormatProperty(Property):
2003    arg_types = {"this": True}
2004
2005
2006class RowFormatDelimitedProperty(Property):
2007    # https://cwiki.apache.org/confluence/display/hive/languagemanual+dml
2008    arg_types = {
2009        "fields": False,
2010        "escaped": False,
2011        "collection_items": False,
2012        "map_keys": False,
2013        "lines": False,
2014        "null": False,
2015        "serde": False,
2016    }
2017
2018
2019class RowFormatSerdeProperty(Property):
2020    arg_types = {"this": True}
2021
2022
2023class SchemaCommentProperty(Property):
2024    arg_types = {"this": True}
2025
2026
2027class SerdeProperties(Property):
2028    arg_types = {"expressions": True}
2029
2030
2031class SetProperty(Property):
2032    arg_types = {"multi": True}
2033
2034
2035class SettingsProperty(Property):
2036    arg_types = {"expressions": True}
2037
2038
2039class SortKeyProperty(Property):
2040    arg_types = {"this": True, "compound": False}
2041
2042
2043class SqlSecurityProperty(Property):
2044    arg_types = {"definer": True}
2045
2046
2047class StabilityProperty(Property):
2048    arg_types = {"this": True}
2049
2050
2051class TemporaryProperty(Property):
2052    arg_types = {}
2053
2054
2055class TransientProperty(Property):
2056    arg_types = {"this": False}
2057
2058
2059class VolatileProperty(Property):
2060    arg_types = {"this": False}
2061
2062
2063class WithDataProperty(Property):
2064    arg_types = {"no": True, "statistics": False}
2065
2066
2067class WithJournalTableProperty(Property):
2068    arg_types = {"this": True}
2069
2070
2071class Properties(Expression):
2072    arg_types = {"expressions": True}
2073
2074    NAME_TO_PROPERTY = {
2075        "ALGORITHM": AlgorithmProperty,
2076        "AUTO_INCREMENT": AutoIncrementProperty,
2077        "CHARACTER SET": CharacterSetProperty,
2078        "CLUSTERED_BY": ClusteredByProperty,
2079        "COLLATE": CollateProperty,
2080        "COMMENT": SchemaCommentProperty,
2081        "DEFINER": DefinerProperty,
2082        "DISTKEY": DistKeyProperty,
2083        "DISTSTYLE": DistStyleProperty,
2084        "ENGINE": EngineProperty,
2085        "EXECUTE AS": ExecuteAsProperty,
2086        "FORMAT": FileFormatProperty,
2087        "LANGUAGE": LanguageProperty,
2088        "LOCATION": LocationProperty,
2089        "PARTITIONED_BY": PartitionedByProperty,
2090        "RETURNS": ReturnsProperty,
2091        "ROW_FORMAT": RowFormatProperty,
2092        "SORTKEY": SortKeyProperty,
2093    }
2094
2095    PROPERTY_TO_NAME = {v: k for k, v in NAME_TO_PROPERTY.items()}
2096
2097    # CREATE property locations
2098    # Form: schema specified
2099    #   create [POST_CREATE]
2100    #     table a [POST_NAME]
2101    #     (b int) [POST_SCHEMA]
2102    #     with ([POST_WITH])
2103    #     index (b) [POST_INDEX]
2104    #
2105    # Form: alias selection
2106    #   create [POST_CREATE]
2107    #     table a [POST_NAME]
2108    #     as [POST_ALIAS] (select * from b) [POST_EXPRESSION]
2109    #     index (c) [POST_INDEX]
2110    class Location(AutoName):
2111        POST_CREATE = auto()
2112        POST_NAME = auto()
2113        POST_SCHEMA = auto()
2114        POST_WITH = auto()
2115        POST_ALIAS = auto()
2116        POST_EXPRESSION = auto()
2117        POST_INDEX = auto()
2118        UNSUPPORTED = auto()
2119
2120    @classmethod
2121    def from_dict(cls, properties_dict: t.Dict) -> Properties:
2122        expressions = []
2123        for key, value in properties_dict.items():
2124            property_cls = cls.NAME_TO_PROPERTY.get(key.upper())
2125            if property_cls:
2126                expressions.append(property_cls(this=convert(value)))
2127            else:
2128                expressions.append(Property(this=Literal.string(key), value=convert(value)))
2129
2130        return cls(expressions=expressions)
2131
2132
2133class Qualify(Expression):
2134    pass
2135
2136
2137# https://www.ibm.com/docs/en/ias?topic=procedures-return-statement-in-sql
2138class Return(Expression):
2139    pass
2140
2141
2142class Reference(Expression):
2143    arg_types = {"this": True, "expressions": False, "options": False}
2144
2145
2146class Tuple(Expression):
2147    arg_types = {"expressions": False}
2148
2149    def isin(
2150        self, *expressions: t.Any, query: t.Optional[ExpOrStr] = None, copy: bool = True, **opts
2151    ) -> In:
2152        return In(
2153            this=_maybe_copy(self, copy),
2154            expressions=[convert(e, copy=copy) for e in expressions],
2155            query=maybe_parse(query, copy=copy, **opts) if query else None,
2156        )
2157
2158
2159class Subqueryable(Unionable):
2160    def subquery(self, alias: t.Optional[ExpOrStr] = None, copy: bool = True) -> Subquery:
2161        """
2162        Convert this expression to an aliased expression that can be used as a Subquery.
2163
2164        Example:
2165            >>> subquery = Select().select("x").from_("tbl").subquery()
2166            >>> Select().select("x").from_(subquery).sql()
2167            'SELECT x FROM (SELECT x FROM tbl)'
2168
2169        Args:
2170            alias (str | Identifier): an optional alias for the subquery
2171            copy (bool): if `False`, modify this expression instance in-place.
2172
2173        Returns:
2174            Alias: the subquery
2175        """
2176        instance = _maybe_copy(self, copy)
2177        if not isinstance(alias, Expression):
2178            alias = TableAlias(this=to_identifier(alias)) if alias else None
2179
2180        return Subquery(this=instance, alias=alias)
2181
2182    def limit(
2183        self, expression: ExpOrStr | int, dialect: DialectType = None, copy: bool = True, **opts
2184    ) -> Select:
2185        raise NotImplementedError
2186
2187    @property
2188    def ctes(self):
2189        with_ = self.args.get("with")
2190        if not with_:
2191            return []
2192        return with_.expressions
2193
2194    @property
2195    def selects(self):
2196        raise NotImplementedError("Subqueryable objects must implement `selects`")
2197
2198    @property
2199    def named_selects(self):
2200        raise NotImplementedError("Subqueryable objects must implement `named_selects`")
2201
2202    def with_(
2203        self,
2204        alias: ExpOrStr,
2205        as_: ExpOrStr,
2206        recursive: t.Optional[bool] = None,
2207        append: bool = True,
2208        dialect: DialectType = None,
2209        copy: bool = True,
2210        **opts,
2211    ) -> Subqueryable:
2212        """
2213        Append to or set the common table expressions.
2214
2215        Example:
2216            >>> Select().with_("tbl2", as_="SELECT * FROM tbl").select("x").from_("tbl2").sql()
2217            'WITH tbl2 AS (SELECT * FROM tbl) SELECT x FROM tbl2'
2218
2219        Args:
2220            alias: the SQL code string to parse as the table name.
2221                If an `Expression` instance is passed, this is used as-is.
2222            as_: the SQL code string to parse as the table expression.
2223                If an `Expression` instance is passed, it will be used as-is.
2224            recursive: set the RECURSIVE part of the expression. Defaults to `False`.
2225            append: if `True`, add to any existing expressions.
2226                Otherwise, this resets the expressions.
2227            dialect: the dialect used to parse the input expression.
2228            copy: if `False`, modify this expression instance in-place.
2229            opts: other options to use to parse the input expressions.
2230
2231        Returns:
2232            The modified expression.
2233        """
2234        return _apply_cte_builder(
2235            self, alias, as_, recursive=recursive, append=append, dialect=dialect, copy=copy, **opts
2236        )
2237
2238
2239QUERY_MODIFIERS = {
2240    "match": False,
2241    "laterals": False,
2242    "joins": False,
2243    "pivots": False,
2244    "where": False,
2245    "group": False,
2246    "having": False,
2247    "qualify": False,
2248    "windows": False,
2249    "distribute": False,
2250    "sort": False,
2251    "cluster": False,
2252    "order": False,
2253    "limit": False,
2254    "offset": False,
2255    "locks": False,
2256    "sample": False,
2257    "settings": False,
2258    "format": False,
2259}
2260
2261
2262# https://learn.microsoft.com/en-us/sql/t-sql/queries/hints-transact-sql-table?view=sql-server-ver16
2263class WithTableHint(Expression):
2264    arg_types = {"expressions": True}
2265
2266
2267# https://dev.mysql.com/doc/refman/8.0/en/index-hints.html
2268class IndexTableHint(Expression):
2269    arg_types = {"this": True, "expressions": False, "target": False}
2270
2271
2272class Table(Expression):
2273    arg_types = {
2274        "this": True,
2275        "alias": False,
2276        "db": False,
2277        "catalog": False,
2278        "laterals": False,
2279        "joins": False,
2280        "pivots": False,
2281        "hints": False,
2282        "system_time": False,
2283    }
2284
2285    @property
2286    def db(self) -> str:
2287        return self.text("db")
2288
2289    @property
2290    def catalog(self) -> str:
2291        return self.text("catalog")
2292
2293    @property
2294    def parts(self) -> t.List[Identifier]:
2295        """Return the parts of a table in order catalog, db, table."""
2296        return [
2297            t.cast(Identifier, self.args[part])
2298            for part in ("catalog", "db", "this")
2299            if self.args.get(part)
2300        ]
2301
2302
2303# See the TSQL "Querying data in a system-versioned temporal table" page
2304class SystemTime(Expression):
2305    arg_types = {
2306        "this": False,
2307        "expression": False,
2308        "kind": True,
2309    }
2310
2311
2312class Union(Subqueryable):
2313    arg_types = {
2314        "with": False,
2315        "this": True,
2316        "expression": True,
2317        "distinct": False,
2318        **QUERY_MODIFIERS,
2319    }
2320
2321    def limit(
2322        self, expression: ExpOrStr | int, dialect: DialectType = None, copy: bool = True, **opts
2323    ) -> Select:
2324        """
2325        Set the LIMIT expression.
2326
2327        Example:
2328            >>> select("1").union(select("1")).limit(1).sql()
2329            'SELECT * FROM (SELECT 1 UNION SELECT 1) AS _l_0 LIMIT 1'
2330
2331        Args:
2332            expression: the SQL code string to parse.
2333                This can also be an integer.
2334                If a `Limit` instance is passed, this is used as-is.
2335                If another `Expression` instance is passed, it will be wrapped in a `Limit`.
2336            dialect: the dialect used to parse the input expression.
2337            copy: if `False`, modify this expression instance in-place.
2338            opts: other options to use to parse the input expressions.
2339
2340        Returns:
2341            The limited subqueryable.
2342        """
2343        return (
2344            select("*")
2345            .from_(self.subquery(alias="_l_0", copy=copy))
2346            .limit(expression, dialect=dialect, copy=False, **opts)
2347        )
2348
2349    def select(
2350        self,
2351        *expressions: t.Optional[ExpOrStr],
2352        append: bool = True,
2353        dialect: DialectType = None,
2354        copy: bool = True,
2355        **opts,
2356    ) -> Union:
2357        """Append to or set the SELECT of the union recursively.
2358
2359        Example:
2360            >>> from sqlglot import parse_one
2361            >>> parse_one("select a from x union select a from y union select a from z").select("b").sql()
2362            'SELECT a, b FROM x UNION SELECT a, b FROM y UNION SELECT a, b FROM z'
2363
2364        Args:
2365            *expressions: the SQL code strings to parse.
2366                If an `Expression` instance is passed, it will be used as-is.
2367            append: if `True`, add to any existing expressions.
2368                Otherwise, this resets the expressions.
2369            dialect: the dialect used to parse the input expressions.
2370            copy: if `False`, modify this expression instance in-place.
2371            opts: other options to use to parse the input expressions.
2372
2373        Returns:
2374            Union: the modified expression.
2375        """
2376        this = self.copy() if copy else self
2377        this.this.unnest().select(*expressions, append=append, dialect=dialect, copy=False, **opts)
2378        this.expression.unnest().select(
2379            *expressions, append=append, dialect=dialect, copy=False, **opts
2380        )
2381        return this
2382
2383    @property
2384    def named_selects(self):
2385        return self.this.unnest().named_selects
2386
2387    @property
2388    def is_star(self) -> bool:
2389        return self.this.is_star or self.expression.is_star
2390
2391    @property
2392    def selects(self):
2393        return self.this.unnest().selects
2394
2395    @property
2396    def left(self):
2397        return self.this
2398
2399    @property
2400    def right(self):
2401        return self.expression
2402
2403
2404class Except(Union):
2405    pass
2406
2407
2408class Intersect(Union):
2409    pass
2410
2411
2412class Unnest(UDTF):
2413    arg_types = {
2414        "expressions": True,
2415        "ordinality": False,
2416        "alias": False,
2417        "offset": False,
2418    }
2419
2420
2421class Update(Expression):
2422    arg_types = {
2423        "with": False,
2424        "this": False,
2425        "expressions": True,
2426        "from": False,
2427        "where": False,
2428        "returning": False,
2429        "limit": False,
2430    }
2431
2432
2433class Values(UDTF):
2434    arg_types = {
2435        "expressions": True,
2436        "ordinality": False,
2437        "alias": False,
2438    }
2439
2440
2441class Var(Expression):
2442    pass
2443
2444
2445class Schema(Expression):
2446    arg_types = {"this": False, "expressions": False}
2447
2448
2449# https://dev.mysql.com/doc/refman/8.0/en/select.html
2450# https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/SELECT.html
2451class Lock(Expression):
2452    arg_types = {"update": True, "expressions": False, "wait": False}
2453
2454
2455class Select(Subqueryable):
2456    arg_types = {
2457        "with": False,
2458        "kind": False,
2459        "expressions": False,
2460        "hint": False,
2461        "distinct": False,
2462        "into": False,
2463        "from": False,
2464        **QUERY_MODIFIERS,
2465    }
2466
2467    def from_(
2468        self, expression: ExpOrStr, dialect: DialectType = None, copy: bool = True, **opts
2469    ) -> Select:
2470        """
2471        Set the FROM expression.
2472
2473        Example:
2474            >>> Select().from_("tbl").select("x").sql()
2475            'SELECT x FROM tbl'
2476
2477        Args:
2478            expression : the SQL code strings to parse.
2479                If a `From` instance is passed, this is used as-is.
2480                If another `Expression` instance is passed, it will be wrapped in a `From`.
2481            dialect: the dialect used to parse the input expression.
2482            copy: if `False`, modify this expression instance in-place.
2483            opts: other options to use to parse the input expressions.
2484
2485        Returns:
2486            The modified Select expression.
2487        """
2488        return _apply_builder(
2489            expression=expression,
2490            instance=self,
2491            arg="from",
2492            into=From,
2493            prefix="FROM",
2494            dialect=dialect,
2495            copy=copy,
2496            **opts,
2497        )
2498
2499    def group_by(
2500        self,
2501        *expressions: t.Optional[ExpOrStr],
2502        append: bool = True,
2503        dialect: DialectType = None,
2504        copy: bool = True,
2505        **opts,
2506    ) -> Select:
2507        """
2508        Set the GROUP BY expression.
2509
2510        Example:
2511            >>> Select().from_("tbl").select("x", "COUNT(1)").group_by("x").sql()
2512            'SELECT x, COUNT(1) FROM tbl GROUP BY x'
2513
2514        Args:
2515            *expressions: the SQL code strings to parse.
2516                If a `Group` instance is passed, this is used as-is.
2517                If another `Expression` instance is passed, it will be wrapped in a `Group`.
2518                If nothing is passed in then a group by is not applied to the expression
2519            append: if `True`, add to any existing expressions.
2520                Otherwise, this flattens all the `Group` expression into a single expression.
2521            dialect: the dialect used to parse the input expression.
2522            copy: if `False`, modify this expression instance in-place.
2523            opts: other options to use to parse the input expressions.
2524
2525        Returns:
2526            The modified Select expression.
2527        """
2528        if not expressions:
2529            return self if not copy else self.copy()
2530
2531        return _apply_child_list_builder(
2532            *expressions,
2533            instance=self,
2534            arg="group",
2535            append=append,
2536            copy=copy,
2537            prefix="GROUP BY",
2538            into=Group,
2539            dialect=dialect,
2540            **opts,
2541        )
2542
2543    def order_by(
2544        self,
2545        *expressions: t.Optional[ExpOrStr],
2546        append: bool = True,
2547        dialect: DialectType = None,
2548        copy: bool = True,
2549        **opts,
2550    ) -> Select:
2551        """
2552        Set the ORDER BY expression.
2553
2554        Example:
2555            >>> Select().from_("tbl").select("x").order_by("x DESC").sql()
2556            'SELECT x FROM tbl ORDER BY x DESC'
2557
2558        Args:
2559            *expressions: the SQL code strings to parse.
2560                If a `Group` instance is passed, this is used as-is.
2561                If another `Expression` instance is passed, it will be wrapped in a `Order`.
2562            append: if `True`, add to any existing expressions.
2563                Otherwise, this flattens all the `Order` expression into a single expression.
2564            dialect: the dialect used to parse the input expression.
2565            copy: if `False`, modify this expression instance in-place.
2566            opts: other options to use to parse the input expressions.
2567
2568        Returns:
2569            The modified Select expression.
2570        """
2571        return _apply_child_list_builder(
2572            *expressions,
2573            instance=self,
2574            arg="order",
2575            append=append,
2576            copy=copy,
2577            prefix="ORDER BY",
2578            into=Order,
2579            dialect=dialect,
2580            **opts,
2581        )
2582
2583    def sort_by(
2584        self,
2585        *expressions: t.Optional[ExpOrStr],
2586        append: bool = True,
2587        dialect: DialectType = None,
2588        copy: bool = True,
2589        **opts,
2590    ) -> Select:
2591        """
2592        Set the SORT BY expression.
2593
2594        Example:
2595            >>> Select().from_("tbl").select("x").sort_by("x DESC").sql(dialect="hive")
2596            'SELECT x FROM tbl SORT BY x DESC'
2597
2598        Args:
2599            *expressions: the SQL code strings to parse.
2600                If a `Group` instance is passed, this is used as-is.
2601                If another `Expression` instance is passed, it will be wrapped in a `SORT`.
2602            append: if `True`, add to any existing expressions.
2603                Otherwise, this flattens all the `Order` expression into a single expression.
2604            dialect: the dialect used to parse the input expression.
2605            copy: if `False`, modify this expression instance in-place.
2606            opts: other options to use to parse the input expressions.
2607
2608        Returns:
2609            The modified Select expression.
2610        """
2611        return _apply_child_list_builder(
2612            *expressions,
2613            instance=self,
2614            arg="sort",
2615            append=append,
2616            copy=copy,
2617            prefix="SORT BY",
2618            into=Sort,
2619            dialect=dialect,
2620            **opts,
2621        )
2622
2623    def cluster_by(
2624        self,
2625        *expressions: t.Optional[ExpOrStr],
2626        append: bool = True,
2627        dialect: DialectType = None,
2628        copy: bool = True,
2629        **opts,
2630    ) -> Select:
2631        """
2632        Set the CLUSTER BY expression.
2633
2634        Example:
2635            >>> Select().from_("tbl").select("x").cluster_by("x DESC").sql(dialect="hive")
2636            'SELECT x FROM tbl CLUSTER BY x DESC'
2637
2638        Args:
2639            *expressions: the SQL code strings to parse.
2640                If a `Group` instance is passed, this is used as-is.
2641                If another `Expression` instance is passed, it will be wrapped in a `Cluster`.
2642            append: if `True`, add to any existing expressions.
2643                Otherwise, this flattens all the `Order` expression into a single expression.
2644            dialect: the dialect used to parse the input expression.
2645            copy: if `False`, modify this expression instance in-place.
2646            opts: other options to use to parse the input expressions.
2647
2648        Returns:
2649            The modified Select expression.
2650        """
2651        return _apply_child_list_builder(
2652            *expressions,
2653            instance=self,
2654            arg="cluster",
2655            append=append,
2656            copy=copy,
2657            prefix="CLUSTER BY",
2658            into=Cluster,
2659            dialect=dialect,
2660            **opts,
2661        )
2662
2663    def limit(
2664        self, expression: ExpOrStr | int, dialect: DialectType = None, copy: bool = True, **opts
2665    ) -> Select:
2666        """
2667        Set the LIMIT expression.
2668
2669        Example:
2670            >>> Select().from_("tbl").select("x").limit(10).sql()
2671            'SELECT x FROM tbl LIMIT 10'
2672
2673        Args:
2674            expression: the SQL code string to parse.
2675                This can also be an integer.
2676                If a `Limit` instance is passed, this is used as-is.
2677                If another `Expression` instance is passed, it will be wrapped in a `Limit`.
2678            dialect: the dialect used to parse the input expression.
2679            copy: if `False`, modify this expression instance in-place.
2680            opts: other options to use to parse the input expressions.
2681
2682        Returns:
2683            Select: the modified expression.
2684        """
2685        return _apply_builder(
2686            expression=expression,
2687            instance=self,
2688            arg="limit",
2689            into=Limit,
2690            prefix="LIMIT",
2691            dialect=dialect,
2692            copy=copy,
2693            **opts,
2694        )
2695
2696    def offset(
2697        self, expression: ExpOrStr | int, dialect: DialectType = None, copy: bool = True, **opts
2698    ) -> Select:
2699        """
2700        Set the OFFSET expression.
2701
2702        Example:
2703            >>> Select().from_("tbl").select("x").offset(10).sql()
2704            'SELECT x FROM tbl OFFSET 10'
2705
2706        Args:
2707            expression: the SQL code string to parse.
2708                This can also be an integer.
2709                If a `Offset` instance is passed, this is used as-is.
2710                If another `Expression` instance is passed, it will be wrapped in a `Offset`.
2711            dialect: the dialect used to parse the input expression.
2712            copy: if `False`, modify this expression instance in-place.
2713            opts: other options to use to parse the input expressions.
2714
2715        Returns:
2716            The modified Select expression.
2717        """
2718        return _apply_builder(
2719            expression=expression,
2720            instance=self,
2721            arg="offset",
2722            into=Offset,
2723            prefix="OFFSET",
2724            dialect=dialect,
2725            copy=copy,
2726            **opts,
2727        )
2728
2729    def select(
2730        self,
2731        *expressions: t.Optional[ExpOrStr],
2732        append: bool = True,
2733        dialect: DialectType = None,
2734        copy: bool = True,
2735        **opts,
2736    ) -> Select:
2737        """
2738        Append to or set the SELECT expressions.
2739
2740        Example:
2741            >>> Select().select("x", "y").sql()
2742            'SELECT x, y'
2743
2744        Args:
2745            *expressions: the SQL code strings to parse.
2746                If an `Expression` instance is passed, it will be used as-is.
2747            append: if `True`, add to any existing expressions.
2748                Otherwise, this resets the expressions.
2749            dialect: the dialect used to parse the input expressions.
2750            copy: if `False`, modify this expression instance in-place.
2751            opts: other options to use to parse the input expressions.
2752
2753        Returns:
2754            The modified Select expression.
2755        """
2756        return _apply_list_builder(
2757            *expressions,
2758            instance=self,
2759            arg="expressions",
2760            append=append,
2761            dialect=dialect,
2762            copy=copy,
2763            **opts,
2764        )
2765
2766    def lateral(
2767        self,
2768        *expressions: t.Optional[ExpOrStr],
2769        append: bool = True,
2770        dialect: DialectType = None,
2771        copy: bool = True,
2772        **opts,
2773    ) -> Select:
2774        """
2775        Append to or set the LATERAL expressions.
2776
2777        Example:
2778            >>> Select().select("x").lateral("OUTER explode(y) tbl2 AS z").from_("tbl").sql()
2779            'SELECT x FROM tbl LATERAL VIEW OUTER EXPLODE(y) tbl2 AS z'
2780
2781        Args:
2782            *expressions: the SQL code strings to parse.
2783                If an `Expression` instance is passed, it will be used as-is.
2784            append: if `True`, add to any existing expressions.
2785                Otherwise, this resets the expressions.
2786            dialect: the dialect used to parse the input expressions.
2787            copy: if `False`, modify this expression instance in-place.
2788            opts: other options to use to parse the input expressions.
2789
2790        Returns:
2791            The modified Select expression.
2792        """
2793        return _apply_list_builder(
2794            *expressions,
2795            instance=self,
2796            arg="laterals",
2797            append=append,
2798            into=Lateral,
2799            prefix="LATERAL VIEW",
2800            dialect=dialect,
2801            copy=copy,
2802            **opts,
2803        )
2804
2805    def join(
2806        self,
2807        expression: ExpOrStr,
2808        on: t.Optional[ExpOrStr] = None,
2809        using: t.Optional[ExpOrStr | t.List[ExpOrStr]] = None,
2810        append: bool = True,
2811        join_type: t.Optional[str] = None,
2812        join_alias: t.Optional[Identifier | str] = None,
2813        dialect: DialectType = None,
2814        copy: bool = True,
2815        **opts,
2816    ) -> Select:
2817        """
2818        Append to or set the JOIN expressions.
2819
2820        Example:
2821            >>> Select().select("*").from_("tbl").join("tbl2", on="tbl1.y = tbl2.y").sql()
2822            'SELECT * FROM tbl JOIN tbl2 ON tbl1.y = tbl2.y'
2823
2824            >>> Select().select("1").from_("a").join("b", using=["x", "y", "z"]).sql()
2825            'SELECT 1 FROM a JOIN b USING (x, y, z)'
2826
2827            Use `join_type` to change the type of join:
2828
2829            >>> Select().select("*").from_("tbl").join("tbl2", on="tbl1.y = tbl2.y", join_type="left outer").sql()
2830            'SELECT * FROM tbl LEFT OUTER JOIN tbl2 ON tbl1.y = tbl2.y'
2831
2832        Args:
2833            expression: the SQL code string to parse.
2834                If an `Expression` instance is passed, it will be used as-is.
2835            on: optionally specify the join "on" criteria as a SQL string.
2836                If an `Expression` instance is passed, it will be used as-is.
2837            using: optionally specify the join "using" criteria as a SQL string.
2838                If an `Expression` instance is passed, it will be used as-is.
2839            append: if `True`, add to any existing expressions.
2840                Otherwise, this resets the expressions.
2841            join_type: if set, alter the parsed join type.
2842            join_alias: an optional alias for the joined source.
2843            dialect: the dialect used to parse the input expressions.
2844            copy: if `False`, modify this expression instance in-place.
2845            opts: other options to use to parse the input expressions.
2846
2847        Returns:
2848            Select: the modified expression.
2849        """
2850        parse_args: t.Dict[str, t.Any] = {"dialect": dialect, **opts}
2851
2852        try:
2853            expression = maybe_parse(expression, into=Join, prefix="JOIN", **parse_args)
2854        except ParseError:
2855            expression = maybe_parse(expression, into=(Join, Expression), **parse_args)
2856
2857        join = expression if isinstance(expression, Join) else Join(this=expression)
2858
2859        if isinstance(join.this, Select):
2860            join.this.replace(join.this.subquery())
2861
2862        if join_type:
2863            method: t.Optional[Token]
2864            side: t.Optional[Token]
2865            kind: t.Optional[Token]
2866
2867            method, side, kind = maybe_parse(join_type, into="JOIN_TYPE", **parse_args)  # type: ignore
2868
2869            if method:
2870                join.set("method", method.text)
2871            if side:
2872                join.set("side", side.text)
2873            if kind:
2874                join.set("kind", kind.text)
2875
2876        if on:
2877            on = and_(*ensure_list(on), dialect=dialect, copy=copy, **opts)
2878            join.set("on", on)
2879
2880        if using:
2881            join = _apply_list_builder(
2882                *ensure_list(using),
2883                instance=join,
2884                arg="using",
2885                append=append,
2886                copy=copy,
2887                **opts,
2888            )
2889
2890        if join_alias:
2891            join.set("this", alias_(join.this, join_alias, table=True))
2892
2893        return _apply_list_builder(
2894            join,
2895            instance=self,
2896            arg="joins",
2897            append=append,
2898            copy=copy,
2899            **opts,
2900        )
2901
2902    def where(
2903        self,
2904        *expressions: t.Optional[ExpOrStr],
2905        append: bool = True,
2906        dialect: DialectType = None,
2907        copy: bool = True,
2908        **opts,
2909    ) -> Select:
2910        """
2911        Append to or set the WHERE expressions.
2912
2913        Example:
2914            >>> Select().select("x").from_("tbl").where("x = 'a' OR x < 'b'").sql()
2915            "SELECT x FROM tbl WHERE x = 'a' OR x < 'b'"
2916
2917        Args:
2918            *expressions: the SQL code strings to parse.
2919                If an `Expression` instance is passed, it will be used as-is.
2920                Multiple expressions are combined with an AND operator.
2921            append: if `True`, AND the new expressions to any existing expression.
2922                Otherwise, this resets the expression.
2923            dialect: the dialect used to parse the input expressions.
2924            copy: if `False`, modify this expression instance in-place.
2925            opts: other options to use to parse the input expressions.
2926
2927        Returns:
2928            Select: the modified expression.
2929        """
2930        return _apply_conjunction_builder(
2931            *expressions,
2932            instance=self,
2933            arg="where",
2934            append=append,
2935            into=Where,
2936            dialect=dialect,
2937            copy=copy,
2938            **opts,
2939        )
2940
2941    def having(
2942        self,
2943        *expressions: t.Optional[ExpOrStr],
2944        append: bool = True,
2945        dialect: DialectType = None,
2946        copy: bool = True,
2947        **opts,
2948    ) -> Select:
2949        """
2950        Append to or set the HAVING expressions.
2951
2952        Example:
2953            >>> Select().select("x", "COUNT(y)").from_("tbl").group_by("x").having("COUNT(y) > 3").sql()
2954            'SELECT x, COUNT(y) FROM tbl GROUP BY x HAVING COUNT(y) > 3'
2955
2956        Args:
2957            *expressions: the SQL code strings to parse.
2958                If an `Expression` instance is passed, it will be used as-is.
2959                Multiple expressions are combined with an AND operator.
2960            append: if `True`, AND the new expressions to any existing expression.
2961                Otherwise, this resets the expression.
2962            dialect: the dialect used to parse the input expressions.
2963            copy: if `False`, modify this expression instance in-place.
2964            opts: other options to use to parse the input expressions.
2965
2966        Returns:
2967            The modified Select expression.
2968        """
2969        return _apply_conjunction_builder(
2970            *expressions,
2971            instance=self,
2972            arg="having",
2973            append=append,
2974            into=Having,
2975            dialect=dialect,
2976            copy=copy,
2977            **opts,
2978        )
2979
2980    def window(
2981        self,
2982        *expressions: t.Optional[ExpOrStr],
2983        append: bool = True,
2984        dialect: DialectType = None,
2985        copy: bool = True,
2986        **opts,
2987    ) -> Select:
2988        return _apply_list_builder(
2989            *expressions,
2990            instance=self,
2991            arg="windows",
2992            append=append,
2993            into=Window,
2994            dialect=dialect,
2995            copy=copy,
2996            **opts,
2997        )
2998
2999    def qualify(
3000        self,
3001        *expressions: t.Optional[ExpOrStr],
3002        append: bool = True,
3003        dialect: DialectType = None,
3004        copy: bool = True,
3005        **opts,
3006    ) -> Select:
3007        return _apply_conjunction_builder(
3008            *expressions,
3009            instance=self,
3010            arg="qualify",
3011            append=append,
3012            into=Qualify,
3013            dialect=dialect,
3014            copy=copy,
3015            **opts,
3016        )
3017
3018    def distinct(
3019        self, *ons: t.Optional[ExpOrStr], distinct: bool = True, copy: bool = True
3020    ) -> Select:
3021        """
3022        Set the OFFSET expression.
3023
3024        Example:
3025            >>> Select().from_("tbl").select("x").distinct().sql()
3026            'SELECT DISTINCT x FROM tbl'
3027
3028        Args:
3029            ons: the expressions to distinct on
3030            distinct: whether the Select should be distinct
3031            copy: if `False`, modify this expression instance in-place.
3032
3033        Returns:
3034            Select: the modified expression.
3035        """
3036        instance = _maybe_copy(self, copy)
3037        on = Tuple(expressions=[maybe_parse(on, copy=copy) for on in ons if on]) if ons else None
3038        instance.set("distinct", Distinct(on=on) if distinct else None)
3039        return instance
3040
3041    def ctas(
3042        self,
3043        table: ExpOrStr,
3044        properties: t.Optional[t.Dict] = None,
3045        dialect: DialectType = None,
3046        copy: bool = True,
3047        **opts,
3048    ) -> Create:
3049        """
3050        Convert this expression to a CREATE TABLE AS statement.
3051
3052        Example:
3053            >>> Select().select("*").from_("tbl").ctas("x").sql()
3054            'CREATE TABLE x AS SELECT * FROM tbl'
3055
3056        Args:
3057            table: the SQL code string to parse as the table name.
3058                If another `Expression` instance is passed, it will be used as-is.
3059            properties: an optional mapping of table properties
3060            dialect: the dialect used to parse the input table.
3061            copy: if `False`, modify this expression instance in-place.
3062            opts: other options to use to parse the input table.
3063
3064        Returns:
3065            The new Create expression.
3066        """
3067        instance = _maybe_copy(self, copy)
3068        table_expression = maybe_parse(
3069            table,
3070            into=Table,
3071            dialect=dialect,
3072            **opts,
3073        )
3074        properties_expression = None
3075        if properties:
3076            properties_expression = Properties.from_dict(properties)
3077
3078        return Create(
3079            this=table_expression,
3080            kind="table",
3081            expression=instance,
3082            properties=properties_expression,
3083        )
3084
3085    def lock(self, update: bool = True, copy: bool = True) -> Select:
3086        """
3087        Set the locking read mode for this expression.
3088
3089        Examples:
3090            >>> Select().select("x").from_("tbl").where("x = 'a'").lock().sql("mysql")
3091            "SELECT x FROM tbl WHERE x = 'a' FOR UPDATE"
3092
3093            >>> Select().select("x").from_("tbl").where("x = 'a'").lock(update=False).sql("mysql")
3094            "SELECT x FROM tbl WHERE x = 'a' FOR SHARE"
3095
3096        Args:
3097            update: if `True`, the locking type will be `FOR UPDATE`, else it will be `FOR SHARE`.
3098            copy: if `False`, modify this expression instance in-place.
3099
3100        Returns:
3101            The modified expression.
3102        """
3103        inst = _maybe_copy(self, copy)
3104        inst.set("locks", [Lock(update=update)])
3105
3106        return inst
3107
3108    def hint(self, *hints: ExpOrStr, dialect: DialectType = None, copy: bool = True) -> Select:
3109        """
3110        Set hints for this expression.
3111
3112        Examples:
3113            >>> Select().select("x").from_("tbl").hint("BROADCAST(y)").sql(dialect="spark")
3114            'SELECT /*+ BROADCAST(y) */ x FROM tbl'
3115
3116        Args:
3117            hints: The SQL code strings to parse as the hints.
3118                If an `Expression` instance is passed, it will be used as-is.
3119            dialect: The dialect used to parse the hints.
3120            copy: If `False`, modify this expression instance in-place.
3121
3122        Returns:
3123            The modified expression.
3124        """
3125        inst = _maybe_copy(self, copy)
3126        inst.set(
3127            "hint", Hint(expressions=[maybe_parse(h, copy=copy, dialect=dialect) for h in hints])
3128        )
3129
3130        return inst
3131
3132    @property
3133    def named_selects(self) -> t.List[str]:
3134        return [e.output_name for e in self.expressions if e.alias_or_name]
3135
3136    @property
3137    def is_star(self) -> bool:
3138        return any(expression.is_star for expression in self.expressions)
3139
3140    @property
3141    def selects(self) -> t.List[Expression]:
3142        return self.expressions
3143
3144
3145class Subquery(DerivedTable, Unionable):
3146    arg_types = {
3147        "this": True,
3148        "alias": False,
3149        "with": False,
3150        **QUERY_MODIFIERS,
3151    }
3152
3153    def unnest(self):
3154        """
3155        Returns the first non subquery.
3156        """
3157        expression = self
3158        while isinstance(expression, Subquery):
3159            expression = expression.this
3160        return expression
3161
3162    @property
3163    def is_star(self) -> bool:
3164        return self.this.is_star
3165
3166    @property
3167    def output_name(self) -> str:
3168        return self.alias
3169
3170
3171class TableSample(Expression):
3172    arg_types = {
3173        "this": False,
3174        "method": False,
3175        "bucket_numerator": False,
3176        "bucket_denominator": False,
3177        "bucket_field": False,
3178        "percent": False,
3179        "rows": False,
3180        "size": False,
3181        "seed": False,
3182        "kind": False,
3183    }
3184
3185
3186class Tag(Expression):
3187    """Tags are used for generating arbitrary sql like SELECT <span>x</span>."""
3188
3189    arg_types = {
3190        "this": False,
3191        "prefix": False,
3192        "postfix": False,
3193    }
3194
3195
3196# Represents both the standard SQL PIVOT operator and DuckDB's "simplified" PIVOT syntax
3197# https://duckdb.org/docs/sql/statements/pivot
3198class Pivot(Expression):
3199    arg_types = {
3200        "this": False,
3201        "alias": False,
3202        "expressions": True,
3203        "field": False,
3204        "unpivot": False,
3205        "using": False,
3206        "group": False,
3207        "columns": False,
3208    }
3209
3210
3211class Window(Expression):
3212    arg_types = {
3213        "this": True,
3214        "partition_by": False,
3215        "order": False,
3216        "spec": False,
3217        "alias": False,
3218        "over": False,
3219        "first": False,
3220    }
3221
3222
3223class WindowSpec(Expression):
3224    arg_types = {
3225        "kind": False,
3226        "start": False,
3227        "start_side": False,
3228        "end": False,
3229        "end_side": False,
3230    }
3231
3232
3233class Where(Expression):
3234    pass
3235
3236
3237class Star(Expression):
3238    arg_types = {"except": False, "replace": False}
3239
3240    @property
3241    def name(self) -> str:
3242        return "*"
3243
3244    @property
3245    def output_name(self) -> str:
3246        return self.name
3247
3248
3249class Parameter(Condition):
3250    arg_types = {"this": True, "wrapped": False}
3251
3252
3253class SessionParameter(Condition):
3254    arg_types = {"this": True, "kind": False}
3255
3256
3257class Placeholder(Condition):
3258    arg_types = {"this": False, "kind": False}
3259
3260
3261class Null(Condition):
3262    arg_types: t.Dict[str, t.Any] = {}
3263
3264    @property
3265    def name(self) -> str:
3266        return "NULL"
3267
3268
3269class Boolean(Condition):
3270    pass
3271
3272
3273class DataTypeSize(Expression):
3274    arg_types = {"this": True, "expression": False}
3275
3276
3277class DataType(Expression):
3278    arg_types = {
3279        "this": True,
3280        "expressions": False,
3281        "nested": False,
3282        "values": False,
3283        "prefix": False,
3284    }
3285
3286    class Type(AutoName):
3287        ARRAY = auto()
3288        BIGDECIMAL = auto()
3289        BIGINT = auto()
3290        BIGSERIAL = auto()
3291        BINARY = auto()
3292        BIT = auto()
3293        BOOLEAN = auto()
3294        CHAR = auto()
3295        DATE = auto()
3296        DATETIME = auto()
3297        DATETIME64 = auto()
3298        ENUM = auto()
3299        INT4RANGE = auto()
3300        INT4MULTIRANGE = auto()
3301        INT8RANGE = auto()
3302        INT8MULTIRANGE = auto()
3303        NUMRANGE = auto()
3304        NUMMULTIRANGE = auto()
3305        TSRANGE = auto()
3306        TSMULTIRANGE = auto()
3307        TSTZRANGE = auto()
3308        TSTZMULTIRANGE = auto()
3309        DATERANGE = auto()
3310        DATEMULTIRANGE = auto()
3311        DECIMAL = auto()
3312        DOUBLE = auto()
3313        FLOAT = auto()
3314        GEOGRAPHY = auto()
3315        GEOMETRY = auto()
3316        HLLSKETCH = auto()
3317        HSTORE = auto()
3318        IMAGE = auto()
3319        INET = auto()
3320        INT = auto()
3321        INT128 = auto()
3322        INT256 = auto()
3323        INTERVAL = auto()
3324        JSON = auto()
3325        JSONB = auto()
3326        LONGBLOB = auto()
3327        LONGTEXT = auto()
3328        MAP = auto()
3329        MEDIUMBLOB = auto()
3330        MEDIUMTEXT = auto()
3331        MONEY = auto()
3332        NCHAR = auto()
3333        NULL = auto()
3334        NULLABLE = auto()
3335        NVARCHAR = auto()
3336        OBJECT = auto()
3337        ROWVERSION = auto()
3338        SERIAL = auto()
3339        SET = auto()
3340        SMALLINT = auto()
3341        SMALLMONEY = auto()
3342        SMALLSERIAL = auto()
3343        STRUCT = auto()
3344        SUPER = auto()
3345        TEXT = auto()
3346        TIME = auto()
3347        TIMESTAMP = auto()
3348        TIMESTAMPTZ = auto()
3349        TIMESTAMPLTZ = auto()
3350        TINYINT = auto()
3351        UBIGINT = auto()
3352        UINT = auto()
3353        USMALLINT = auto()
3354        UTINYINT = auto()
3355        UNKNOWN = auto()  # Sentinel value, useful for type annotation
3356        UINT128 = auto()
3357        UINT256 = auto()
3358        UNIQUEIDENTIFIER = auto()
3359        USERDEFINED = "USER-DEFINED"
3360        UUID = auto()
3361        VARBINARY = auto()
3362        VARCHAR = auto()
3363        VARIANT = auto()
3364        XML = auto()
3365
3366    TEXT_TYPES = {
3367        Type.CHAR,
3368        Type.NCHAR,
3369        Type.VARCHAR,
3370        Type.NVARCHAR,
3371        Type.TEXT,
3372    }
3373
3374    INTEGER_TYPES = {
3375        Type.INT,
3376        Type.TINYINT,
3377        Type.SMALLINT,
3378        Type.BIGINT,
3379        Type.INT128,
3380        Type.INT256,
3381    }
3382
3383    FLOAT_TYPES = {
3384        Type.FLOAT,
3385        Type.DOUBLE,
3386    }
3387
3388    NUMERIC_TYPES = {*INTEGER_TYPES, *FLOAT_TYPES}
3389
3390    TEMPORAL_TYPES = {
3391        Type.TIME,
3392        Type.TIMESTAMP,
3393        Type.TIMESTAMPTZ,
3394        Type.TIMESTAMPLTZ,
3395        Type.DATE,
3396        Type.DATETIME,
3397        Type.DATETIME64,
3398    }
3399
3400    META_TYPES = {"UNKNOWN", "NULL"}
3401
3402    @classmethod
3403    def build(
3404        cls, dtype: str | DataType | DataType.Type, dialect: DialectType = None, **kwargs
3405    ) -> DataType:
3406        from sqlglot import parse_one
3407
3408        if isinstance(dtype, str):
3409            upper = dtype.upper()
3410            if upper in DataType.META_TYPES:
3411                data_type_exp: t.Optional[Expression] = DataType(this=DataType.Type[upper])
3412            else:
3413                data_type_exp = parse_one(dtype, read=dialect, into=DataType)
3414
3415            if data_type_exp is None:
3416                raise ValueError(f"Unparsable data type value: {dtype}")
3417        elif isinstance(dtype, DataType.Type):
3418            data_type_exp = DataType(this=dtype)
3419        elif isinstance(dtype, DataType):
3420            return dtype
3421        else:
3422            raise ValueError(f"Invalid data type: {type(dtype)}. Expected str or DataType.Type")
3423
3424        return DataType(**{**data_type_exp.args, **kwargs})
3425
3426    def is_type(self, *dtypes: str | DataType | DataType.Type) -> bool:
3427        return any(self.this == DataType.build(dtype).this for dtype in dtypes)
3428
3429
3430# https://www.postgresql.org/docs/15/datatype-pseudo.html
3431class PseudoType(Expression):
3432    pass
3433
3434
3435# WHERE x <OP> EXISTS|ALL|ANY|SOME(SELECT ...)
3436class SubqueryPredicate(Predicate):
3437    pass
3438
3439
3440class All(SubqueryPredicate):
3441    pass
3442
3443
3444class Any(SubqueryPredicate):
3445    pass
3446
3447
3448class Exists(SubqueryPredicate):
3449    pass
3450
3451
3452# Commands to interact with the databases or engines. For most of the command
3453# expressions we parse whatever comes after the command's name as a string.
3454class Command(Expression):
3455    arg_types = {"this": True, "expression": False}
3456
3457
3458class Transaction(Expression):
3459    arg_types = {"this": False, "modes": False}
3460
3461
3462class Commit(Expression):
3463    arg_types = {"chain": False}
3464
3465
3466class Rollback(Expression):
3467    arg_types = {"savepoint": False}
3468
3469
3470class AlterTable(Expression):
3471    arg_types = {"this": True, "actions": True, "exists": False}
3472
3473
3474class AddConstraint(Expression):
3475    arg_types = {"this": False, "expression": False, "enforced": False}
3476
3477
3478class DropPartition(Expression):
3479    arg_types = {"expressions": True, "exists": False}
3480
3481
3482# Binary expressions like (ADD a b)
3483class Binary(Condition):
3484    arg_types = {"this": True, "expression": True}
3485
3486    @property
3487    def left(self):
3488        return self.this
3489
3490    @property
3491    def right(self):
3492        return self.expression
3493
3494
3495class Add(Binary):
3496    pass
3497
3498
3499class Connector(Binary):
3500    pass
3501
3502
3503class And(Connector):
3504    pass
3505
3506
3507class Or(Connector):
3508    pass
3509
3510
3511class BitwiseAnd(Binary):
3512    pass
3513
3514
3515class BitwiseLeftShift(Binary):
3516    pass
3517
3518
3519class BitwiseOr(Binary):
3520    pass
3521
3522
3523class BitwiseRightShift(Binary):
3524    pass
3525
3526
3527class BitwiseXor(Binary):
3528    pass
3529
3530
3531class Div(Binary):
3532    pass
3533
3534
3535class Overlaps(Binary):
3536    pass
3537
3538
3539class Dot(Binary):
3540    @property
3541    def name(self) -> str:
3542        return self.expression.name
3543
3544    @property
3545    def output_name(self) -> str:
3546        return self.name
3547
3548    @classmethod
3549    def build(self, expressions: t.Sequence[Expression]) -> Dot:
3550        """Build a Dot object with a sequence of expressions."""
3551        if len(expressions) < 2:
3552            raise ValueError(f"Dot requires >= 2 expressions.")
3553
3554        a, b, *expressions = expressions
3555        dot = Dot(this=a, expression=b)
3556
3557        for expression in expressions:
3558            dot = Dot(this=dot, expression=expression)
3559
3560        return dot
3561
3562
3563class DPipe(Binary):
3564    pass
3565
3566
3567class SafeDPipe(DPipe):
3568    pass
3569
3570
3571class EQ(Binary, Predicate):
3572    pass
3573
3574
3575class NullSafeEQ(Binary, Predicate):
3576    pass
3577
3578
3579class NullSafeNEQ(Binary, Predicate):
3580    pass
3581
3582
3583class Distance(Binary):
3584    pass
3585
3586
3587class Escape(Binary):
3588    pass
3589
3590
3591class Glob(Binary, Predicate):
3592    pass
3593
3594
3595class GT(Binary, Predicate):
3596    pass
3597
3598
3599class GTE(Binary, Predicate):
3600    pass
3601
3602
3603class ILike(Binary, Predicate):
3604    pass
3605
3606
3607class ILikeAny(Binary, Predicate):
3608    pass
3609
3610
3611class IntDiv(Binary):
3612    pass
3613
3614
3615class Is(Binary, Predicate):
3616    pass
3617
3618
3619class Kwarg(Binary):
3620    """Kwarg in special functions like func(kwarg => y)."""
3621
3622
3623class Like(Binary, Predicate):
3624    pass
3625
3626
3627class LikeAny(Binary, Predicate):
3628    pass
3629
3630
3631class LT(Binary, Predicate):
3632    pass
3633
3634
3635class LTE(Binary, Predicate):
3636    pass
3637
3638
3639class Mod(Binary):
3640    pass
3641
3642
3643class Mul(Binary):
3644    pass
3645
3646
3647class NEQ(Binary, Predicate):
3648    pass
3649
3650
3651class SimilarTo(Binary, Predicate):
3652    pass
3653
3654
3655class Slice(Binary):
3656    arg_types = {"this": False, "expression": False}
3657
3658
3659class Sub(Binary):
3660    pass
3661
3662
3663class ArrayOverlaps(Binary):
3664    pass
3665
3666
3667# Unary Expressions
3668# (NOT a)
3669class Unary(Condition):
3670    pass
3671
3672
3673class BitwiseNot(Unary):
3674    pass
3675
3676
3677class Not(Unary):
3678    pass
3679
3680
3681class Paren(Unary):
3682    arg_types = {"this": True, "with": False}
3683
3684    @property
3685    def output_name(self) -> str:
3686        return self.this.name
3687
3688
3689class Neg(Unary):
3690    pass
3691
3692
3693class Alias(Expression):
3694    arg_types = {"this": True, "alias": False}
3695
3696    @property
3697    def output_name(self) -> str:
3698        return self.alias
3699
3700
3701class Aliases(Expression):
3702    arg_types = {"this": True, "expressions": True}
3703
3704    @property
3705    def aliases(self):
3706        return self.expressions
3707
3708
3709class AtTimeZone(Expression):
3710    arg_types = {"this": True, "zone": True}
3711
3712
3713class Between(Predicate):
3714    arg_types = {"this": True, "low": True, "high": True}
3715
3716
3717class Bracket(Condition):
3718    arg_types = {"this": True, "expressions": True}
3719
3720
3721class SafeBracket(Bracket):
3722    """Represents array lookup where OOB index yields NULL instead of causing a failure."""
3723
3724
3725class Distinct(Expression):
3726    arg_types = {"expressions": False, "on": False}
3727
3728
3729class In(Predicate):
3730    arg_types = {
3731        "this": True,
3732        "expressions": False,
3733        "query": False,
3734        "unnest": False,
3735        "field": False,
3736        "is_global": False,
3737    }
3738
3739
3740class TimeUnit(Expression):
3741    """Automatically converts unit arg into a var."""
3742
3743    arg_types = {"unit": False}
3744
3745    def __init__(self, **args):
3746        unit = args.get("unit")
3747        if isinstance(unit, (Column, Literal)):
3748            args["unit"] = Var(this=unit.name)
3749        elif isinstance(unit, Week):
3750            unit.set("this", Var(this=unit.this.name))
3751
3752        super().__init__(**args)
3753
3754
3755class Interval(TimeUnit):
3756    arg_types = {"this": False, "unit": False}
3757
3758    @property
3759    def unit(self) -> t.Optional[Var]:
3760        return self.args.get("unit")
3761
3762
3763class IgnoreNulls(Expression):
3764    pass
3765
3766
3767class RespectNulls(Expression):
3768    pass
3769
3770
3771# Functions
3772class Func(Condition):
3773    """
3774    The base class for all function expressions.
3775
3776    Attributes:
3777        is_var_len_args (bool): if set to True the last argument defined in arg_types will be
3778            treated as a variable length argument and the argument's value will be stored as a list.
3779        _sql_names (list): determines the SQL name (1st item in the list) and aliases (subsequent items)
3780            for this function expression. These values are used to map this node to a name during parsing
3781            as well as to provide the function's name during SQL string generation. By default the SQL
3782            name is set to the expression's class name transformed to snake case.
3783    """
3784
3785    is_var_len_args = False
3786
3787    @classmethod
3788    def from_arg_list(cls, args):
3789        if cls.is_var_len_args:
3790            all_arg_keys = list(cls.arg_types)
3791            # If this function supports variable length argument treat the last argument as such.
3792            non_var_len_arg_keys = all_arg_keys[:-1] if cls.is_var_len_args else all_arg_keys
3793            num_non_var = len(non_var_len_arg_keys)
3794
3795            args_dict = {arg_key: arg for arg, arg_key in zip(args, non_var_len_arg_keys)}
3796            args_dict[all_arg_keys[-1]] = args[num_non_var:]
3797        else:
3798            args_dict = {arg_key: arg for arg, arg_key in zip(args, cls.arg_types)}
3799
3800        return cls(**args_dict)
3801
3802    @classmethod
3803    def sql_names(cls):
3804        if cls is Func:
3805            raise NotImplementedError(
3806                "SQL name is only supported by concrete function implementations"
3807            )
3808        if "_sql_names" not in cls.__dict__:
3809            cls._sql_names = [camel_to_snake_case(cls.__name__)]
3810        return cls._sql_names
3811
3812    @classmethod
3813    def sql_name(cls):
3814        return cls.sql_names()[0]
3815
3816    @classmethod
3817    def default_parser_mappings(cls):
3818        return {name: cls.from_arg_list for name in cls.sql_names()}
3819
3820
3821class AggFunc(Func):
3822    pass
3823
3824
3825class ParameterizedAgg(AggFunc):
3826    arg_types = {"this": True, "expressions": True, "params": True}
3827
3828
3829class Abs(Func):
3830    pass
3831
3832
3833class Anonymous(Func):
3834    arg_types = {"this": True, "expressions": False}
3835    is_var_len_args = True
3836
3837
3838# https://docs.snowflake.com/en/sql-reference/functions/hll
3839# https://docs.aws.amazon.com/redshift/latest/dg/r_HLL_function.html
3840class Hll(AggFunc):
3841    arg_types = {"this": True, "expressions": False}
3842    is_var_len_args = True
3843
3844
3845class ApproxDistinct(AggFunc):
3846    arg_types = {"this": True, "accuracy": False}
3847    _sql_names = ["APPROX_DISTINCT", "APPROX_COUNT_DISTINCT"]
3848
3849
3850class Array(Func):
3851    arg_types = {"expressions": False}
3852    is_var_len_args = True
3853
3854
3855# https://docs.snowflake.com/en/sql-reference/functions/to_char
3856class ToChar(Func):
3857    arg_types = {"this": True, "format": False}
3858
3859
3860class GenerateSeries(Func):
3861    arg_types = {"start": True, "end": True, "step": False}
3862
3863
3864class ArrayAgg(AggFunc):
3865    pass
3866
3867
3868class ArrayAll(Func):
3869    arg_types = {"this": True, "expression": True}
3870
3871
3872class ArrayAny(Func):
3873    arg_types = {"this": True, "expression": True}
3874
3875
3876class ArrayConcat(Func):
3877    arg_types = {"this": True, "expressions": False}
3878    is_var_len_args = True
3879
3880
3881class ArrayContains(Binary, Func):
3882    pass
3883
3884
3885class ArrayContained(Binary):
3886    pass
3887
3888
3889class ArrayFilter(Func):
3890    arg_types = {"this": True, "expression": True}
3891    _sql_names = ["FILTER", "ARRAY_FILTER"]
3892
3893
3894class ArrayJoin(Func):
3895    arg_types = {"this": True, "expression": True, "null": False}
3896
3897
3898class ArraySize(Func):
3899    arg_types = {"this": True, "expression": False}
3900
3901
3902class ArraySort(Func):
3903    arg_types = {"this": True, "expression": False}
3904
3905
3906class ArraySum(Func):
3907    pass
3908
3909
3910class ArrayUnionAgg(AggFunc):
3911    pass
3912
3913
3914class Avg(AggFunc):
3915    pass
3916
3917
3918class AnyValue(AggFunc):
3919    pass
3920
3921
3922class Case(Func):
3923    arg_types = {"this": False, "ifs": True, "default": False}
3924
3925    def when(self, condition: ExpOrStr, then: ExpOrStr, copy: bool = True, **opts) -> Case:
3926        instance = _maybe_copy(self, copy)
3927        instance.append(
3928            "ifs",
3929            If(
3930                this=maybe_parse(condition, copy=copy, **opts),
3931                true=maybe_parse(then, copy=copy, **opts),
3932            ),
3933        )
3934        return instance
3935
3936    def else_(self, condition: ExpOrStr, copy: bool = True, **opts) -> Case:
3937        instance = _maybe_copy(self, copy)
3938        instance.set("default", maybe_parse(condition, copy=copy, **opts))
3939        return instance
3940
3941
3942class Cast(Func):
3943    arg_types = {"this": True, "to": True}
3944
3945    @property
3946    def name(self) -> str:
3947        return self.this.name
3948
3949    @property
3950    def to(self) -> DataType:
3951        return self.args["to"]
3952
3953    @property
3954    def output_name(self) -> str:
3955        return self.name
3956
3957    def is_type(self, *dtypes: str | DataType | DataType.Type) -> bool:
3958        return self.to.is_type(*dtypes)
3959
3960
3961class CastToStrType(Func):
3962    arg_types = {"this": True, "expression": True}
3963
3964
3965class Collate(Binary):
3966    pass
3967
3968
3969class TryCast(Cast):
3970    pass
3971
3972
3973class Ceil(Func):
3974    arg_types = {"this": True, "decimals": False}
3975    _sql_names = ["CEIL", "CEILING"]
3976
3977
3978class Coalesce(Func):
3979    arg_types = {"this": True, "expressions": False}
3980    is_var_len_args = True
3981    _sql_names = ["COALESCE", "IFNULL", "NVL"]
3982
3983
3984class Concat(Func):
3985    arg_types = {"expressions": True}
3986    is_var_len_args = True
3987
3988
3989class SafeConcat(Concat):
3990    pass
3991
3992
3993class ConcatWs(Concat):
3994    _sql_names = ["CONCAT_WS"]
3995
3996
3997class Count(AggFunc):
3998    arg_types = {"this": False, "expressions": False}
3999    is_var_len_args = True
4000
4001
4002class CountIf(AggFunc):
4003    pass
4004
4005
4006class CurrentDate(Func):
4007    arg_types = {"this": False}
4008
4009
4010class CurrentDatetime(Func):
4011    arg_types = {"this": False}
4012
4013
4014class CurrentTime(Func):
4015    arg_types = {"this": False}
4016
4017
4018class CurrentTimestamp(Func):
4019    arg_types = {"this": False}
4020
4021
4022class CurrentUser(Func):
4023    arg_types = {"this": False}
4024
4025
4026class DateAdd(Func, TimeUnit):
4027    arg_types = {"this": True, "expression": True, "unit": False}
4028
4029
4030class DateSub(Func, TimeUnit):
4031    arg_types = {"this": True, "expression": True, "unit": False}
4032
4033
4034class DateDiff(Func, TimeUnit):
4035    _sql_names = ["DATEDIFF", "DATE_DIFF"]
4036    arg_types = {"this": True, "expression": True, "unit": False}
4037
4038
4039class DateTrunc(Func):
4040    arg_types = {"unit": True, "this": True, "zone": False}
4041
4042
4043class DatetimeAdd(Func, TimeUnit):
4044    arg_types = {"this": True, "expression": True, "unit": False}
4045
4046
4047class DatetimeSub(Func, TimeUnit):
4048    arg_types = {"this": True, "expression": True, "unit": False}
4049
4050
4051class DatetimeDiff(Func, TimeUnit):
4052    arg_types = {"this": True, "expression": True, "unit": False}
4053
4054
4055class DatetimeTrunc(Func, TimeUnit):
4056    arg_types = {"this": True, "unit": True, "zone": False}
4057
4058
4059class DayOfWeek(Func):
4060    _sql_names = ["DAY_OF_WEEK", "DAYOFWEEK"]
4061
4062
4063class DayOfMonth(Func):
4064    _sql_names = ["DAY_OF_MONTH", "DAYOFMONTH"]
4065
4066
4067class DayOfYear(Func):
4068    _sql_names = ["DAY_OF_YEAR", "DAYOFYEAR"]
4069
4070
4071class WeekOfYear(Func):
4072    _sql_names = ["WEEK_OF_YEAR", "WEEKOFYEAR"]
4073
4074
4075class LastDateOfMonth(Func):
4076    pass
4077
4078
4079class Extract(Func):
4080    arg_types = {"this": True, "expression": True}
4081
4082
4083class TimestampAdd(Func, TimeUnit):
4084    arg_types = {"this": True, "expression": True, "unit": False}
4085
4086
4087class TimestampSub(Func, TimeUnit):
4088    arg_types = {"this": True, "expression": True, "unit": False}
4089
4090
4091class TimestampDiff(Func, TimeUnit):
4092    arg_types = {"this": True, "expression": True, "unit": False}
4093
4094
4095class TimestampTrunc(Func, TimeUnit):
4096    arg_types = {"this": True, "unit": True, "zone": False}
4097
4098
4099class TimeAdd(Func, TimeUnit):
4100    arg_types = {"this": True, "expression": True, "unit": False}
4101
4102
4103class TimeSub(Func, TimeUnit):
4104    arg_types = {"this": True, "expression": True, "unit": False}
4105
4106
4107class TimeDiff(Func, TimeUnit):
4108    arg_types = {"this": True, "expression": True, "unit": False}
4109
4110
4111class TimeTrunc(Func, TimeUnit):
4112    arg_types = {"this": True, "unit": True, "zone": False}
4113
4114
4115class DateFromParts(Func):
4116    _sql_names = ["DATEFROMPARTS"]
4117    arg_types = {"year": True, "month": True, "day": True}
4118
4119
4120class DateStrToDate(Func):
4121    pass
4122
4123
4124class DateToDateStr(Func):
4125    pass
4126
4127
4128class DateToDi(Func):
4129    pass
4130
4131
4132class Date(Func):
4133    arg_types = {"expressions": True}
4134    is_var_len_args = True
4135
4136
4137class Day(Func):
4138    pass
4139
4140
4141class Decode(Func):
4142    arg_types = {"this": True, "charset": True, "replace": False}
4143
4144
4145class DiToDate(Func):
4146    pass
4147
4148
4149class Encode(Func):
4150    arg_types = {"this": True, "charset": True}
4151
4152
4153class Exp(Func):
4154    pass
4155
4156
4157class Explode(Func):
4158    pass
4159
4160
4161class Floor(Func):
4162    arg_types = {"this": True, "decimals": False}
4163
4164
4165class FromBase64(Func):
4166    pass
4167
4168
4169class ToBase64(Func):
4170    pass
4171
4172
4173class Greatest(Func):
4174    arg_types = {"this": True, "expressions": False}
4175    is_var_len_args = True
4176
4177
4178class GroupConcat(Func):
4179    arg_types = {"this": True, "separator": False}
4180
4181
4182class Hex(Func):
4183    pass
4184
4185
4186class If(Func):
4187    arg_types = {"this": True, "true": True, "false": False}
4188
4189
4190class Initcap(Func):
4191    arg_types = {"this": True, "expression": False}
4192
4193
4194class JSONKeyValue(Expression):
4195    arg_types = {"this": True, "expression": True}
4196
4197
4198class JSONObject(Func):
4199    arg_types = {
4200        "expressions": False,
4201        "null_handling": False,
4202        "unique_keys": False,
4203        "return_type": False,
4204        "format_json": False,
4205        "encoding": False,
4206    }
4207
4208
4209class OpenJSONColumnDef(Expression):
4210    arg_types = {"this": True, "kind": True, "path": False, "as_json": False}
4211
4212
4213class OpenJSON(Func):
4214    arg_types = {"this": True, "path": False, "expressions": False}
4215
4216
4217class JSONBContains(Binary):
4218    _sql_names = ["JSONB_CONTAINS"]
4219
4220
4221class JSONExtract(Binary, Func):
4222    _sql_names = ["JSON_EXTRACT"]
4223
4224
4225class JSONExtractScalar(JSONExtract):
4226    _sql_names = ["JSON_EXTRACT_SCALAR"]
4227
4228
4229class JSONBExtract(JSONExtract):
4230    _sql_names = ["JSONB_EXTRACT"]
4231
4232
4233class JSONBExtractScalar(JSONExtract):
4234    _sql_names = ["JSONB_EXTRACT_SCALAR"]
4235
4236
4237class JSONFormat(Func):
4238    arg_types = {"this": False, "options": False}
4239    _sql_names = ["JSON_FORMAT"]
4240
4241
4242class Least(Func):
4243    arg_types = {"expressions": False}
4244    is_var_len_args = True
4245
4246
4247class Left(Func):
4248    arg_types = {"this": True, "expression": True}
4249
4250
4251class Right(Func):
4252    arg_types = {"this": True, "expression": True}
4253
4254
4255class Length(Func):
4256    _sql_names = ["LENGTH", "LEN"]
4257
4258
4259class Levenshtein(Func):
4260    arg_types = {
4261        "this": True,
4262        "expression": False,
4263        "ins_cost": False,
4264        "del_cost": False,
4265        "sub_cost": False,
4266    }
4267
4268
4269class Ln(Func):
4270    pass
4271
4272
4273class Log(Func):
4274    arg_types = {"this": True, "expression": False}
4275
4276
4277class Log2(Func):
4278    pass
4279
4280
4281class Log10(Func):
4282    pass
4283
4284
4285class LogicalOr(AggFunc):
4286    _sql_names = ["LOGICAL_OR", "BOOL_OR", "BOOLOR_AGG"]
4287
4288
4289class LogicalAnd(AggFunc):
4290    _sql_names = ["LOGICAL_AND", "BOOL_AND", "BOOLAND_AGG"]
4291
4292
4293class Lower(Func):
4294    _sql_names = ["LOWER", "LCASE"]
4295
4296
4297class Map(Func):
4298    arg_types = {"keys": False, "values": False}
4299
4300
4301class MapFromEntries(Func):
4302    pass
4303
4304
4305class StarMap(Func):
4306    pass
4307
4308
4309class VarMap(Func):
4310    arg_types = {"keys": True, "values": True}
4311    is_var_len_args = True
4312
4313    @property
4314    def keys(self) -> t.List[Expression]:
4315        return self.args["keys"].expressions
4316
4317    @property
4318    def values(self) -> t.List[Expression]:
4319        return self.args["values"].expressions
4320
4321
4322# https://dev.mysql.com/doc/refman/8.0/en/fulltext-search.html
4323class MatchAgainst(Func):
4324    arg_types = {"this": True, "expressions": True, "modifier": False}
4325
4326
4327class Max(AggFunc):
4328    arg_types = {"this": True, "expressions": False}
4329    is_var_len_args = True
4330
4331
4332class MD5(Func):
4333    _sql_names = ["MD5"]
4334
4335
4336class Min(AggFunc):
4337    arg_types = {"this": True, "expressions": False}
4338    is_var_len_args = True
4339
4340
4341class Month(Func):
4342    pass
4343
4344
4345class Nvl2(Func):
4346    arg_types = {"this": True, "true": True, "false": False}
4347
4348
4349class Posexplode(Func):
4350    pass
4351
4352
4353class Pow(Binary, Func):
4354    _sql_names = ["POWER", "POW"]
4355
4356
4357class PercentileCont(AggFunc):
4358    arg_types = {"this": True, "expression": False}
4359
4360
4361class PercentileDisc(AggFunc):
4362    arg_types = {"this": True, "expression": False}
4363
4364
4365class Quantile(AggFunc):
4366    arg_types = {"this": True, "quantile": True}
4367
4368
4369class ApproxQuantile(Quantile):
4370    arg_types = {"this": True, "quantile": True, "accuracy": False, "weight": False}
4371
4372
4373class RangeN(Func):
4374    arg_types = {"this": True, "expressions": True, "each": False}
4375
4376
4377class ReadCSV(Func):
4378    _sql_names = ["READ_CSV"]
4379    is_var_len_args = True
4380    arg_types = {"this": True, "expressions": False}
4381
4382
4383class Reduce(Func):
4384    arg_types = {"this": True, "initial": True, "merge": True, "finish": False}
4385
4386
4387class RegexpExtract(Func):
4388    arg_types = {
4389        "this": True,
4390        "expression": True,
4391        "position": False,
4392        "occurrence": False,
4393        "group": False,
4394    }
4395
4396
4397class RegexpLike(Func):
4398    arg_types = {"this": True, "expression": True, "flag": False}
4399
4400
4401class RegexpILike(Func):
4402    arg_types = {"this": True, "expression": True, "flag": False}
4403
4404
4405# https://spark.apache.org/docs/latest/api/python/reference/pyspark.sql/api/pyspark.sql.functions.split.html
4406# limit is the number of times a pattern is applied
4407class RegexpSplit(Func):
4408    arg_types = {"this": True, "expression": True, "limit": False}
4409
4410
4411class Repeat(Func):
4412    arg_types = {"this": True, "times": True}
4413
4414
4415class Round(Func):
4416    arg_types = {"this": True, "decimals": False}
4417
4418
4419class RowNumber(Func):
4420    arg_types: t.Dict[str, t.Any] = {}
4421
4422
4423class SafeDivide(Func):
4424    arg_types = {"this": True, "expression": True}
4425
4426
4427class SetAgg(AggFunc):
4428    pass
4429
4430
4431class SHA(Func):
4432    _sql_names = ["SHA", "SHA1"]
4433
4434
4435class SHA2(Func):
4436    _sql_names = ["SHA2"]
4437    arg_types = {"this": True, "length": False}
4438
4439
4440class SortArray(Func):
4441    arg_types = {"this": True, "asc": False}
4442
4443
4444class Split(Func):
4445    arg_types = {"this": True, "expression": True, "limit": False}
4446
4447
4448# Start may be omitted in the case of postgres
4449# https://www.postgresql.org/docs/9.1/functions-string.html @ Table 9-6
4450class Substring(Func):
4451    arg_types = {"this": True, "start": False, "length": False}
4452
4453
4454class StandardHash(Func):
4455    arg_types = {"this": True, "expression": False}
4456
4457
4458class StrPosition(Func):
4459    arg_types = {
4460        "this": True,
4461        "substr": True,
4462        "position": False,
4463        "instance": False,
4464    }
4465
4466
4467class StrToDate(Func):
4468    arg_types = {"this": True, "format": True}
4469
4470
4471class StrToTime(Func):
4472    arg_types = {"this": True, "format": True}
4473
4474
4475# Spark allows unix_timestamp()
4476# https://spark.apache.org/docs/3.1.3/api/python/reference/api/pyspark.sql.functions.unix_timestamp.html
4477class StrToUnix(Func):
4478    arg_types = {"this": False, "format": False}
4479
4480
4481class NumberToStr(Func):
4482    arg_types = {"this": True, "format": True}
4483
4484
4485class FromBase(Func):
4486    arg_types = {"this": True, "expression": True}
4487
4488
4489class Struct(Func):
4490    arg_types = {"expressions": True}
4491    is_var_len_args = True
4492
4493
4494class StructExtract(Func):
4495    arg_types = {"this": True, "expression": True}
4496
4497
4498class Sum(AggFunc):
4499    pass
4500
4501
4502class Sqrt(Func):
4503    pass
4504
4505
4506class Stddev(AggFunc):
4507    pass
4508
4509
4510class StddevPop(AggFunc):
4511    pass
4512
4513
4514class StddevSamp(AggFunc):
4515    pass
4516
4517
4518class TimeToStr(Func):
4519    arg_types = {"this": True, "format": True}
4520
4521
4522class TimeToTimeStr(Func):
4523    pass
4524
4525
4526class TimeToUnix(Func):
4527    pass
4528
4529
4530class TimeStrToDate(Func):
4531    pass
4532
4533
4534class TimeStrToTime(Func):
4535    pass
4536
4537
4538class TimeStrToUnix(Func):
4539    pass
4540
4541
4542class Trim(Func):
4543    arg_types = {
4544        "this": True,
4545        "expression": False,
4546        "position": False,
4547        "collation": False,
4548    }
4549
4550
4551class TsOrDsAdd(Func, TimeUnit):
4552    arg_types = {"this": True, "expression": True, "unit": False}
4553
4554
4555class TsOrDsToDateStr(Func):
4556    pass
4557
4558
4559class TsOrDsToDate(Func):
4560    arg_types = {"this": True, "format": False}
4561
4562
4563class TsOrDiToDi(Func):
4564    pass
4565
4566
4567class Unhex(Func):
4568    pass
4569
4570
4571class UnixToStr(Func):
4572    arg_types = {"this": True, "format": False}
4573
4574
4575# https://prestodb.io/docs/current/functions/datetime.html
4576# presto has weird zone/hours/minutes
4577class UnixToTime(Func):
4578    arg_types = {"this": True, "scale": False, "zone": False, "hours": False, "minutes": False}
4579
4580    SECONDS = Literal.string("seconds")
4581    MILLIS = Literal.string("millis")
4582    MICROS = Literal.string("micros")
4583
4584
4585class UnixToTimeStr(Func):
4586    pass
4587
4588
4589class Upper(Func):
4590    _sql_names = ["UPPER", "UCASE"]
4591
4592
4593class Variance(AggFunc):
4594    _sql_names = ["VARIANCE", "VARIANCE_SAMP", "VAR_SAMP"]
4595
4596
4597class VariancePop(AggFunc):
4598    _sql_names = ["VARIANCE_POP", "VAR_POP"]
4599
4600
4601class Week(Func):
4602    arg_types = {"this": True, "mode": False}
4603
4604
4605class XMLTable(Func):
4606    arg_types = {"this": True, "passing": False, "columns": False, "by_ref": False}
4607
4608
4609class Year(Func):
4610    pass
4611
4612
4613class Use(Expression):
4614    arg_types = {"this": True, "kind": False}
4615
4616
4617class Merge(Expression):
4618    arg_types = {"this": True, "using": True, "on": True, "expressions": True}
4619
4620
4621class When(Func):
4622    arg_types = {"matched": True, "source": False, "condition": False, "then": True}
4623
4624
4625# https://docs.oracle.com/javadb/10.8.3.0/ref/rrefsqljnextvaluefor.html
4626# https://learn.microsoft.com/en-us/sql/t-sql/functions/next-value-for-transact-sql?view=sql-server-ver16
4627class NextValueFor(Func):
4628    arg_types = {"this": True, "order": False}
4629
4630
4631def _norm_arg(arg):
4632    return arg.lower() if type(arg) is str else arg
4633
4634
4635ALL_FUNCTIONS = subclasses(__name__, Func, (AggFunc, Anonymous, Func))
4636
4637
4638# Helpers
4639@t.overload
4640def maybe_parse(
4641    sql_or_expression: ExpOrStr,
4642    *,
4643    into: t.Type[E],
4644    dialect: DialectType = None,
4645    prefix: t.Optional[str] = None,
4646    copy: bool = False,
4647    **opts,
4648) -> E:
4649    ...
4650
4651
4652@t.overload
4653def maybe_parse(
4654    sql_or_expression: str | E,
4655    *,
4656    into: t.Optional[IntoType] = None,
4657    dialect: DialectType = None,
4658    prefix: t.Optional[str] = None,
4659    copy: bool = False,
4660    **opts,
4661) -> E:
4662    ...
4663
4664
4665def maybe_parse(
4666    sql_or_expression: ExpOrStr,
4667    *,
4668    into: t.Optional[IntoType] = None,
4669    dialect: DialectType = None,
4670    prefix: t.Optional[str] = None,
4671    copy: bool = False,
4672    **opts,
4673) -> Expression:
4674    """Gracefully handle a possible string or expression.
4675
4676    Example:
4677        >>> maybe_parse("1")
4678        (LITERAL this: 1, is_string: False)
4679        >>> maybe_parse(to_identifier("x"))
4680        (IDENTIFIER this: x, quoted: False)
4681
4682    Args:
4683        sql_or_expression: the SQL code string or an expression
4684        into: the SQLGlot Expression to parse into
4685        dialect: the dialect used to parse the input expressions (in the case that an
4686            input expression is a SQL string).
4687        prefix: a string to prefix the sql with before it gets parsed
4688            (automatically includes a space)
4689        copy: whether or not to copy the expression.
4690        **opts: other options to use to parse the input expressions (again, in the case
4691            that an input expression is a SQL string).
4692
4693    Returns:
4694        Expression: the parsed or given expression.
4695    """
4696    if isinstance(sql_or_expression, Expression):
4697        if copy:
4698            return sql_or_expression.copy()
4699        return sql_or_expression
4700
4701    if sql_or_expression is None:
4702        raise ParseError(f"SQL cannot be None")
4703
4704    import sqlglot
4705
4706    sql = str(sql_or_expression)
4707    if prefix:
4708        sql = f"{prefix} {sql}"
4709
4710    return sqlglot.parse_one(sql, read=dialect, into=into, **opts)
4711
4712
4713def _maybe_copy(instance: E, copy: bool = True) -> E:
4714    return instance.copy() if copy else instance
4715
4716
4717def _is_wrong_expression(expression, into):
4718    return isinstance(expression, Expression) and not isinstance(expression, into)
4719
4720
4721def _apply_builder(
4722    expression,
4723    instance,
4724    arg,
4725    copy=True,
4726    prefix=None,
4727    into=None,
4728    dialect=None,
4729    **opts,
4730):
4731    if _is_wrong_expression(expression, into):
4732        expression = into(this=expression)
4733    instance = _maybe_copy(instance, copy)
4734    expression = maybe_parse(
4735        sql_or_expression=expression,
4736        prefix=prefix,
4737        into=into,
4738        dialect=dialect,
4739        **opts,
4740    )
4741    instance.set(arg, expression)
4742    return instance
4743
4744
4745def _apply_child_list_builder(
4746    *expressions,
4747    instance,
4748    arg,
4749    append=True,
4750    copy=True,
4751    prefix=None,
4752    into=None,
4753    dialect=None,
4754    properties=None,
4755    **opts,
4756):
4757    instance = _maybe_copy(instance, copy)
4758    parsed = []
4759    for expression in expressions:
4760        if expression is not None:
4761            if _is_wrong_expression(expression, into):
4762                expression = into(expressions=[expression])
4763
4764            expression = maybe_parse(
4765                expression,
4766                into=into,
4767                dialect=dialect,
4768                prefix=prefix,
4769                **opts,
4770            )
4771            parsed.extend(expression.expressions)
4772
4773    existing = instance.args.get(arg)
4774    if append and existing:
4775        parsed = existing.expressions + parsed
4776
4777    child = into(expressions=parsed)
4778    for k, v in (properties or {}).items():
4779        child.set(k, v)
4780    instance.set(arg, child)
4781
4782    return instance
4783
4784
4785def _apply_list_builder(
4786    *expressions,
4787    instance,
4788    arg,
4789    append=True,
4790    copy=True,
4791    prefix=None,
4792    into=None,
4793    dialect=None,
4794    **opts,
4795):
4796    inst = _maybe_copy(instance, copy)
4797
4798    expressions = [
4799        maybe_parse(
4800            sql_or_expression=expression,
4801            into=into,
4802            prefix=prefix,
4803            dialect=dialect,
4804            **opts,
4805        )
4806        for expression in expressions
4807        if expression is not None
4808    ]
4809
4810    existing_expressions = inst.args.get(arg)
4811    if append and existing_expressions:
4812        expressions = existing_expressions + expressions
4813
4814    inst.set(arg, expressions)
4815    return inst
4816
4817
4818def _apply_conjunction_builder(
4819    *expressions,
4820    instance,
4821    arg,
4822    into=None,
4823    append=True,
4824    copy=True,
4825    dialect=None,
4826    **opts,
4827):
4828    expressions = [exp for exp in expressions if exp is not None and exp != ""]
4829    if not expressions:
4830        return instance
4831
4832    inst = _maybe_copy(instance, copy)
4833
4834    existing = inst.args.get(arg)
4835    if append and existing is not None:
4836        expressions = [existing.this if into else existing] + list(expressions)
4837
4838    node = and_(*expressions, dialect=dialect, copy=copy, **opts)
4839
4840    inst.set(arg, into(this=node) if into else node)
4841    return inst
4842
4843
4844def _apply_cte_builder(
4845    instance: E,
4846    alias: ExpOrStr,
4847    as_: ExpOrStr,
4848    recursive: t.Optional[bool] = None,
4849    append: bool = True,
4850    dialect: DialectType = None,
4851    copy: bool = True,
4852    **opts,
4853) -> E:
4854    alias_expression = maybe_parse(alias, dialect=dialect, into=TableAlias, **opts)
4855    as_expression = maybe_parse(as_, dialect=dialect, **opts)
4856    cte = CTE(this=as_expression, alias=alias_expression)
4857    return _apply_child_list_builder(
4858        cte,
4859        instance=instance,
4860        arg="with",
4861        append=append,
4862        copy=copy,
4863        into=With,
4864        properties={"recursive": recursive or False},
4865    )
4866
4867
4868def _combine(
4869    expressions: t.Sequence[t.Optional[ExpOrStr]],
4870    operator: t.Type[Connector],
4871    dialect: DialectType = None,
4872    copy: bool = True,
4873    **opts,
4874) -> Expression:
4875    conditions = [
4876        condition(expression, dialect=dialect, copy=copy, **opts)
4877        for expression in expressions
4878        if expression is not None
4879    ]
4880
4881    this, *rest = conditions
4882    if rest:
4883        this = _wrap(this, Connector)
4884    for expression in rest:
4885        this = operator(this=this, expression=_wrap(expression, Connector))
4886
4887    return this
4888
4889
4890def _wrap(expression: E, kind: t.Type[Expression]) -> E | Paren:
4891    return Paren(this=expression) if isinstance(expression, kind) else expression
4892
4893
4894def union(
4895    left: ExpOrStr, right: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
4896) -> Union:
4897    """
4898    Initializes a syntax tree from one UNION expression.
4899
4900    Example:
4901        >>> union("SELECT * FROM foo", "SELECT * FROM bla").sql()
4902        'SELECT * FROM foo UNION SELECT * FROM bla'
4903
4904    Args:
4905        left: the SQL code string corresponding to the left-hand side.
4906            If an `Expression` instance is passed, it will be used as-is.
4907        right: the SQL code string corresponding to the right-hand side.
4908            If an `Expression` instance is passed, it will be used as-is.
4909        distinct: set the DISTINCT flag if and only if this is true.
4910        dialect: the dialect used to parse the input expression.
4911        opts: other options to use to parse the input expressions.
4912
4913    Returns:
4914        The new Union instance.
4915    """
4916    left = maybe_parse(sql_or_expression=left, dialect=dialect, **opts)
4917    right = maybe_parse(sql_or_expression=right, dialect=dialect, **opts)
4918
4919    return Union(this=left, expression=right, distinct=distinct)
4920
4921
4922def intersect(
4923    left: ExpOrStr, right: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
4924) -> Intersect:
4925    """
4926    Initializes a syntax tree from one INTERSECT expression.
4927
4928    Example:
4929        >>> intersect("SELECT * FROM foo", "SELECT * FROM bla").sql()
4930        'SELECT * FROM foo INTERSECT SELECT * FROM bla'
4931
4932    Args:
4933        left: the SQL code string corresponding to the left-hand side.
4934            If an `Expression` instance is passed, it will be used as-is.
4935        right: the SQL code string corresponding to the right-hand side.
4936            If an `Expression` instance is passed, it will be used as-is.
4937        distinct: set the DISTINCT flag if and only if this is true.
4938        dialect: the dialect used to parse the input expression.
4939        opts: other options to use to parse the input expressions.
4940
4941    Returns:
4942        The new Intersect instance.
4943    """
4944    left = maybe_parse(sql_or_expression=left, dialect=dialect, **opts)
4945    right = maybe_parse(sql_or_expression=right, dialect=dialect, **opts)
4946
4947    return Intersect(this=left, expression=right, distinct=distinct)
4948
4949
4950def except_(
4951    left: ExpOrStr, right: ExpOrStr, distinct: bool = True, dialect: DialectType = None, **opts
4952) -> Except:
4953    """
4954    Initializes a syntax tree from one EXCEPT expression.
4955
4956    Example:
4957        >>> except_("SELECT * FROM foo", "SELECT * FROM bla").sql()
4958        'SELECT * FROM foo EXCEPT SELECT * FROM bla'
4959
4960    Args:
4961        left: the SQL code string corresponding to the left-hand side.
4962            If an `Expression` instance is passed, it will be used as-is.
4963        right: the SQL code string corresponding to the right-hand side.
4964            If an `Expression` instance is passed, it will be used as-is.
4965        distinct: set the DISTINCT flag if and only if this is true.
4966        dialect: the dialect used to parse the input expression.
4967        opts: other options to use to parse the input expressions.
4968
4969    Returns:
4970        The new Except instance.
4971    """
4972    left = maybe_parse(sql_or_expression=left, dialect=dialect, **opts)
4973    right = maybe_parse(sql_or_expression=right, dialect=dialect, **opts)
4974
4975    return Except(this=left, expression=right, distinct=distinct)
4976
4977
4978def select(*expressions: ExpOrStr, dialect: DialectType = None, **opts) -> Select:
4979    """
4980    Initializes a syntax tree from one or multiple SELECT expressions.
4981
4982    Example:
4983        >>> select("col1", "col2").from_("tbl").sql()
4984        'SELECT col1, col2 FROM tbl'
4985
4986    Args:
4987        *expressions: the SQL code string to parse as the expressions of a
4988            SELECT statement. If an Expression instance is passed, this is used as-is.
4989        dialect: the dialect used to parse the input expressions (in the case that an
4990            input expression is a SQL string).
4991        **opts: other options to use to parse the input expressions (again, in the case
4992            that an input expression is a SQL string).
4993
4994    Returns:
4995        Select: the syntax tree for the SELECT statement.
4996    """
4997    return Select().select(*expressions, dialect=dialect, **opts)
4998
4999
5000def from_(expression: ExpOrStr, dialect: DialectType = None, **opts) -> Select:
5001    """
5002    Initializes a syntax tree from a FROM expression.
5003
5004    Example:
5005        >>> from_("tbl").select("col1", "col2").sql()
5006        'SELECT col1, col2 FROM tbl'
5007
5008    Args:
5009        *expression: the SQL code string to parse as the FROM expressions of a
5010            SELECT statement. If an Expression instance is passed, this is used as-is.
5011        dialect: the dialect used to parse the input expression (in the case that the
5012            input expression is a SQL string).
5013        **opts: other options to use to parse the input expressions (again, in the case
5014            that the input expression is a SQL string).
5015
5016    Returns:
5017        Select: the syntax tree for the SELECT statement.
5018    """
5019    return Select().from_(expression, dialect=dialect, **opts)
5020
5021
5022def update(
5023    table: str | Table,
5024    properties: dict,
5025    where: t.Optional[ExpOrStr] = None,
5026    from_: t.Optional[ExpOrStr] = None,
5027    dialect: DialectType = None,
5028    **opts,
5029) -> Update:
5030    """
5031    Creates an update statement.
5032
5033    Example:
5034        >>> update("my_table", {"x": 1, "y": "2", "z": None}, from_="baz", where="id > 1").sql()
5035        "UPDATE my_table SET x = 1, y = '2', z = NULL FROM baz WHERE id > 1"
5036
5037    Args:
5038        *properties: dictionary of properties to set which are
5039            auto converted to sql objects eg None -> NULL
5040        where: sql conditional parsed into a WHERE statement
5041        from_: sql statement parsed into a FROM statement
5042        dialect: the dialect used to parse the input expressions.
5043        **opts: other options to use to parse the input expressions.
5044
5045    Returns:
5046        Update: the syntax tree for the UPDATE statement.
5047    """
5048    update_expr = Update(this=maybe_parse(table, into=Table, dialect=dialect))
5049    update_expr.set(
5050        "expressions",
5051        [
5052            EQ(this=maybe_parse(k, dialect=dialect, **opts), expression=convert(v))
5053            for k, v in properties.items()
5054        ],
5055    )
5056    if from_:
5057        update_expr.set(
5058            "from",
5059            maybe_parse(from_, into=From, dialect=dialect, prefix="FROM", **opts),
5060        )
5061    if isinstance(where, Condition):
5062        where = Where(this=where)
5063    if where:
5064        update_expr.set(
5065            "where",
5066            maybe_parse(where, into=Where, dialect=dialect, prefix="WHERE", **opts),
5067        )
5068    return update_expr
5069
5070
5071def delete(
5072    table: ExpOrStr,
5073    where: t.Optional[ExpOrStr] = None,
5074    returning: t.Optional[ExpOrStr] = None,
5075    dialect: DialectType = None,
5076    **opts,
5077) -> Delete:
5078    """
5079    Builds a delete statement.
5080
5081    Example:
5082        >>> delete("my_table", where="id > 1").sql()
5083        'DELETE FROM my_table WHERE id > 1'
5084
5085    Args:
5086        where: sql conditional parsed into a WHERE statement
5087        returning: sql conditional parsed into a RETURNING statement
5088        dialect: the dialect used to parse the input expressions.
5089        **opts: other options to use to parse the input expressions.
5090
5091    Returns:
5092        Delete: the syntax tree for the DELETE statement.
5093    """
5094    delete_expr = Delete().delete(table, dialect=dialect, copy=False, **opts)
5095    if where:
5096        delete_expr = delete_expr.where(where, dialect=dialect, copy=False, **opts)
5097    if returning:
5098        delete_expr = delete_expr.returning(returning, dialect=dialect, copy=False, **opts)
5099    return delete_expr
5100
5101
5102def insert(
5103    expression: ExpOrStr,
5104    into: ExpOrStr,
5105    columns: t.Optional[t.Sequence[ExpOrStr]] = None,
5106    overwrite: t.Optional[bool] = None,
5107    dialect: DialectType = None,
5108    copy: bool = True,
5109    **opts,
5110) -> Insert:
5111    """
5112    Builds an INSERT statement.
5113
5114    Example:
5115        >>> insert("VALUES (1, 2, 3)", "tbl").sql()
5116        'INSERT INTO tbl VALUES (1, 2, 3)'
5117
5118    Args:
5119        expression: the sql string or expression of the INSERT statement
5120        into: the tbl to insert data to.
5121        columns: optionally the table's column names.
5122        overwrite: whether to INSERT OVERWRITE or not.
5123        dialect: the dialect used to parse the input expressions.
5124        copy: whether or not to copy the expression.
5125        **opts: other options to use to parse the input expressions.
5126
5127    Returns:
5128        Insert: the syntax tree for the INSERT statement.
5129    """
5130    expr = maybe_parse(expression, dialect=dialect, copy=copy, **opts)
5131    this: Table | Schema = maybe_parse(into, into=Table, dialect=dialect, copy=copy, **opts)
5132
5133    if columns:
5134        this = _apply_list_builder(
5135            *columns,
5136            instance=Schema(this=this),
5137            arg="expressions",
5138            into=Identifier,
5139            copy=False,
5140            dialect=dialect,
5141            **opts,
5142        )
5143
5144    return Insert(this=this, expression=expr, overwrite=overwrite)
5145
5146
5147def condition(
5148    expression: ExpOrStr, dialect: DialectType = None, copy: bool = True, **opts
5149) -> Condition:
5150    """
5151    Initialize a logical condition expression.
5152
5153    Example:
5154        >>> condition("x=1").sql()
5155        'x = 1'
5156
5157        This is helpful for composing larger logical syntax trees:
5158        >>> where = condition("x=1")
5159        >>> where = where.and_("y=1")
5160        >>> Select().from_("tbl").select("*").where(where).sql()
5161        'SELECT * FROM tbl WHERE x = 1 AND y = 1'
5162
5163    Args:
5164        *expression: the SQL code string to parse.
5165            If an Expression instance is passed, this is used as-is.
5166        dialect: the dialect used to parse the input expression (in the case that the
5167            input expression is a SQL string).
5168        copy: Whether or not to copy `expression` (only applies to expressions).
5169        **opts: other options to use to parse the input expressions (again, in the case
5170            that the input expression is a SQL string).
5171
5172    Returns:
5173        The new Condition instance
5174    """
5175    return maybe_parse(
5176        expression,
5177        into=Condition,
5178        dialect=dialect,
5179        copy=copy,
5180        **opts,
5181    )
5182
5183
5184def and_(
5185    *expressions: t.Optional[ExpOrStr], dialect: DialectType = None, copy: bool = True, **opts
5186) -> Condition:
5187    """
5188    Combine multiple conditions with an AND logical operator.
5189
5190    Example:
5191        >>> and_("x=1", and_("y=1", "z=1")).sql()
5192        'x = 1 AND (y = 1 AND z = 1)'
5193
5194    Args:
5195        *expressions: the SQL code strings to parse.
5196            If an Expression instance is passed, this is used as-is.
5197        dialect: the dialect used to parse the input expression.
5198        copy: whether or not to copy `expressions` (only applies to Expressions).
5199        **opts: other options to use to parse the input expressions.
5200
5201    Returns:
5202        And: the new condition
5203    """
5204    return t.cast(Condition, _combine(expressions, And, dialect, copy=copy, **opts))
5205
5206
5207def or_(
5208    *expressions: t.Optional[ExpOrStr], dialect: DialectType = None, copy: bool = True, **opts
5209) -> Condition:
5210    """
5211    Combine multiple conditions with an OR logical operator.
5212
5213    Example:
5214        >>> or_("x=1", or_("y=1", "z=1")).sql()
5215        'x = 1 OR (y = 1 OR z = 1)'
5216
5217    Args:
5218        *expressions: the SQL code strings to parse.
5219            If an Expression instance is passed, this is used as-is.
5220        dialect: the dialect used to parse the input expression.
5221        copy: whether or not to copy `expressions` (only applies to Expressions).
5222        **opts: other options to use to parse the input expressions.
5223
5224    Returns:
5225        Or: the new condition
5226    """
5227    return t.cast(Condition, _combine(expressions, Or, dialect, copy=copy, **opts))
5228
5229
5230def not_(expression: ExpOrStr, dialect: DialectType = None, copy: bool = True, **opts) -> Not:
5231    """
5232    Wrap a condition with a NOT operator.
5233
5234    Example:
5235        >>> not_("this_suit='black'").sql()
5236        "NOT this_suit = 'black'"
5237
5238    Args:
5239        expression: the SQL code string to parse.
5240            If an Expression instance is passed, this is used as-is.
5241        dialect: the dialect used to parse the input expression.
5242        copy: whether to copy the expression or not.
5243        **opts: other options to use to parse the input expressions.
5244
5245    Returns:
5246        The new condition.
5247    """
5248    this = condition(
5249        expression,
5250        dialect=dialect,
5251        copy=copy,
5252        **opts,
5253    )
5254    return Not(this=_wrap(this, Connector))
5255
5256
5257def paren(expression: ExpOrStr, copy: bool = True) -> Paren:
5258    """
5259    Wrap an expression in parentheses.
5260
5261    Example:
5262        >>> paren("5 + 3").sql()
5263        '(5 + 3)'
5264
5265    Args:
5266        expression: the SQL code string to parse.
5267            If an Expression instance is passed, this is used as-is.
5268        copy: whether to copy the expression or not.
5269
5270    Returns:
5271        The wrapped expression.
5272    """
5273    return Paren(this=maybe_parse(expression, copy=copy))
5274
5275
5276SAFE_IDENTIFIER_RE = re.compile(r"^[_a-zA-Z][\w]*$")
5277
5278
5279@t.overload
5280def to_identifier(name: None, quoted: t.Optional[bool] = None, copy: bool = True) -> None:
5281    ...
5282
5283
5284@t.overload
5285def to_identifier(
5286    name: str | Identifier, quoted: t.Optional[bool] = None, copy: bool = True
5287) -> Identifier:
5288    ...
5289
5290
5291def to_identifier(name, quoted=None, copy=True):
5292    """Builds an identifier.
5293
5294    Args:
5295        name: The name to turn into an identifier.
5296        quoted: Whether or not force quote the identifier.
5297        copy: Whether or not to copy a passed in Identefier node.
5298
5299    Returns:
5300        The identifier ast node.
5301    """
5302
5303    if name is None:
5304        return None
5305
5306    if isinstance(name, Identifier):
5307        identifier = _maybe_copy(name, copy)
5308    elif isinstance(name, str):
5309        identifier = Identifier(
5310            this=name,
5311            quoted=not SAFE_IDENTIFIER_RE.match(name) if quoted is None else quoted,
5312        )
5313    else:
5314        raise ValueError(f"Name needs to be a string or an Identifier, got: {name.__class__}")
5315    return identifier
5316
5317
5318INTERVAL_STRING_RE = re.compile(r"\s*([0-9]+)\s*([a-zA-Z]+)\s*")
5319
5320
5321def to_interval(interval: str | Literal) -> Interval:
5322    """Builds an interval expression from a string like '1 day' or '5 months'."""
5323    if isinstance(interval, Literal):
5324        if not interval.is_string:
5325            raise ValueError("Invalid interval string.")
5326
5327        interval = interval.this
5328
5329    interval_parts = INTERVAL_STRING_RE.match(interval)  # type: ignore
5330
5331    if not interval_parts:
5332        raise ValueError("Invalid interval string.")
5333
5334    return Interval(
5335        this=Literal.string(interval_parts.group(1)),
5336        unit=Var(this=interval_parts.group(2)),
5337    )
5338
5339
5340@t.overload
5341def to_table(sql_path: str | Table, **kwargs) -> Table:
5342    ...
5343
5344
5345@t.overload
5346def to_table(sql_path: None, **kwargs) -> None:
5347    ...
5348
5349
5350def to_table(
5351    sql_path: t.Optional[str | Table], dialect: DialectType = None, **kwargs
5352) -> t.Optional[Table]:
5353    """
5354    Create a table expression from a `[catalog].[schema].[table]` sql path. Catalog and schema are optional.
5355    If a table is passed in then that table is returned.
5356
5357    Args:
5358        sql_path: a `[catalog].[schema].[table]` string.
5359        dialect: the source dialect according to which the table name will be parsed.
5360        kwargs: the kwargs to instantiate the resulting `Table` expression with.
5361
5362    Returns:
5363        A table expression.
5364    """
5365    if sql_path is None or isinstance(sql_path, Table):
5366        return sql_path
5367    if not isinstance(sql_path, str):
5368        raise ValueError(f"Invalid type provided for a table: {type(sql_path)}")
5369
5370    table = maybe_parse(sql_path, into=Table, dialect=dialect)
5371    if table:
5372        for k, v in kwargs.items():
5373            table.set(k, v)
5374
5375    return table
5376
5377
5378def to_column(sql_path: str | Column, **kwargs) -> Column:
5379    """
5380    Create a column from a `[table].[column]` sql path. Schema is optional.
5381
5382    If a column is passed in then that column is returned.
5383
5384    Args:
5385        sql_path: `[table].[column]` string
5386    Returns:
5387        Table: A column expression
5388    """
5389    if sql_path is None or isinstance(sql_path, Column):
5390        return sql_path
5391    if not isinstance(sql_path, str):
5392        raise ValueError(f"Invalid type provided for column: {type(sql_path)}")
5393    return column(*reversed(sql_path.split(".")), **kwargs)  # type: ignore
5394
5395
5396def alias_(
5397    expression: ExpOrStr,
5398    alias: str | Identifier,
5399    table: bool | t.Sequence[str | Identifier] = False,
5400    quoted: t.Optional[bool] = None,
5401    dialect: DialectType = None,
5402    copy: bool = True,
5403    **opts,
5404):
5405    """Create an Alias expression.
5406
5407    Example:
5408        >>> alias_('foo', 'bar').sql()
5409        'foo AS bar'
5410
5411        >>> alias_('(select 1, 2)', 'bar', table=['a', 'b']).sql()
5412        '(SELECT 1, 2) AS bar(a, b)'
5413
5414    Args:
5415        expression: the SQL code strings to parse.
5416            If an Expression instance is passed, this is used as-is.
5417        alias: the alias name to use. If the name has
5418            special characters it is quoted.
5419        table: Whether or not to create a table alias, can also be a list of columns.
5420        quoted: whether or not to quote the alias
5421        dialect: the dialect used to parse the input expression.
5422        copy: Whether or not to copy the expression.
5423        **opts: other options to use to parse the input expressions.
5424
5425    Returns:
5426        Alias: the aliased expression
5427    """
5428    exp = maybe_parse(expression, dialect=dialect, copy=copy, **opts)
5429    alias = to_identifier(alias, quoted=quoted)
5430
5431    if table:
5432        table_alias = TableAlias(this=alias)
5433        exp.set("alias", table_alias)
5434
5435        if not isinstance(table, bool):
5436            for column in table:
5437                table_alias.append("columns", to_identifier(column, quoted=quoted))
5438
5439        return exp
5440
5441    # We don't set the "alias" arg for Window expressions, because that would add an IDENTIFIER node in
5442    # the AST, representing a "named_window" [1] construct (eg. bigquery). What we want is an ALIAS node
5443    # for the complete Window expression.
5444    #
5445    # [1]: https://cloud.google.com/bigquery/docs/reference/standard-sql/window-function-calls
5446
5447    if "alias" in exp.arg_types and not isinstance(exp, Window):
5448        exp.set("alias", alias)
5449        return exp
5450    return Alias(this=exp, alias=alias)
5451
5452
5453def subquery(
5454    expression: ExpOrStr,
5455    alias: t.Optional[Identifier | str] = None,
5456    dialect: DialectType = None,
5457    **opts,
5458) -> Select:
5459    """
5460    Build a subquery expression.
5461
5462    Example:
5463        >>> subquery('select x from tbl', 'bar').select('x').sql()
5464        'SELECT x FROM (SELECT x FROM tbl) AS bar'
5465
5466    Args:
5467        expression: the SQL code strings to parse.
5468            If an Expression instance is passed, this is used as-is.
5469        alias: the alias name to use.
5470        dialect: the dialect used to parse the input expression.
5471        **opts: other options to use to parse the input expressions.
5472
5473    Returns:
5474        A new Select instance with the subquery expression included.
5475    """
5476
5477    expression = maybe_parse(expression, dialect=dialect, **opts).subquery(alias)
5478    return Select().from_(expression, dialect=dialect, **opts)
5479
5480
5481def column(
5482    col: str | Identifier,
5483    table: t.Optional[str | Identifier] = None,
5484    db: t.Optional[str | Identifier] = None,
5485    catalog: t.Optional[str | Identifier] = None,
5486    quoted: t.Optional[bool] = None,
5487) -> Column:
5488    """
5489    Build a Column.
5490
5491    Args:
5492        col: Column name.
5493        table: Table name.
5494        db: Database name.
5495        catalog: Catalog name.
5496        quoted: Whether to force quotes on the column's identifiers.
5497
5498    Returns:
5499        The new Column instance.
5500    """
5501    return Column(
5502        this=to_identifier(col, quoted=quoted),
5503        table=to_identifier(table, quoted=quoted),
5504        db=to_identifier(db, quoted=quoted),
5505        catalog=to_identifier(catalog, quoted=quoted),
5506    )
5507
5508
5509def cast(expression: ExpOrStr, to: str | DataType | DataType.Type, **opts) -> Cast:
5510    """Cast an expression to a data type.
5511
5512    Example:
5513        >>> cast('x + 1', 'int').sql()
5514        'CAST(x + 1 AS INT)'
5515
5516    Args:
5517        expression: The expression to cast.
5518        to: The datatype to cast to.
5519
5520    Returns:
5521        The new Cast instance.
5522    """
5523    expression = maybe_parse(expression, **opts)
5524    return Cast(this=expression, to=DataType.build(to, **opts))
5525
5526
5527def table_(
5528    table: Identifier | str,
5529    db: t.Optional[Identifier | str] = None,
5530    catalog: t.Optional[Identifier | str] = None,
5531    quoted: t.Optional[bool] = None,
5532    alias: t.Optional[Identifier | str] = None,
5533) -> Table:
5534    """Build a Table.
5535
5536    Args:
5537        table: Table name.
5538        db: Database name.
5539        catalog: Catalog name.
5540        quote: Whether to force quotes on the table's identifiers.
5541        alias: Table's alias.
5542
5543    Returns:
5544        The new Table instance.
5545    """
5546    return Table(
5547        this=to_identifier(table, quoted=quoted),
5548        db=to_identifier(db, quoted=quoted),
5549        catalog=to_identifier(catalog, quoted=quoted),
5550        alias=TableAlias(this=to_identifier(alias)) if alias else None,
5551    )
5552
5553
5554def values(
5555    values: t.Iterable[t.Tuple[t.Any, ...]],
5556    alias: t.Optional[str] = None,
5557    columns: t.Optional[t.Iterable[str] | t.Dict[str, DataType]] = None,
5558) -> Values:
5559    """Build VALUES statement.
5560
5561    Example:
5562        >>> values([(1, '2')]).sql()
5563        "VALUES (1, '2')"
5564
5565    Args:
5566        values: values statements that will be converted to SQL
5567        alias: optional alias
5568        columns: Optional list of ordered column names or ordered dictionary of column names to types.
5569         If either are provided then an alias is also required.
5570
5571    Returns:
5572        Values: the Values expression object
5573    """
5574    if columns and not alias:
5575        raise ValueError("Alias is required when providing columns")
5576
5577    return Values(
5578        expressions=[convert(tup) for tup in values],
5579        alias=(
5580            TableAlias(this=to_identifier(alias), columns=[to_identifier(x) for x in columns])
5581            if columns
5582            else (TableAlias(this=to_identifier(alias)) if alias else None)
5583        ),
5584    )
5585
5586
5587def var(name: t.Optional[ExpOrStr]) -> Var:
5588    """Build a SQL variable.
5589
5590    Example:
5591        >>> repr(var('x'))
5592        '(VAR this: x)'
5593
5594        >>> repr(var(column('x', table='y')))
5595        '(VAR this: x)'
5596
5597    Args:
5598        name: The name of the var or an expression who's name will become the var.
5599
5600    Returns:
5601        The new variable node.
5602    """
5603    if not name:
5604        raise ValueError("Cannot convert empty name into var.")
5605
5606    if isinstance(name, Expression):
5607        name = name.name
5608    return Var(this=name)
5609
5610
5611def rename_table(old_name: str | Table, new_name: str | Table) -> AlterTable:
5612    """Build ALTER TABLE... RENAME... expression
5613
5614    Args:
5615        old_name: The old name of the table
5616        new_name: The new name of the table
5617
5618    Returns:
5619        Alter table expression
5620    """
5621    old_table = to_table(old_name)
5622    new_table = to_table(new_name)
5623    return AlterTable(
5624        this=old_table,
5625        actions=[
5626            RenameTable(this=new_table),
5627        ],
5628    )
5629
5630
5631def convert(value: t.Any, copy: bool = False) -> Expression:
5632    """Convert a python value into an expression object.
5633
5634    Raises an error if a conversion is not possible.
5635
5636    Args:
5637        value: A python object.
5638        copy: Whether or not to copy `value` (only applies to Expressions and collections).
5639
5640    Returns:
5641        Expression: the equivalent expression object.
5642    """
5643    if isinstance(value, Expression):
5644        return _maybe_copy(value, copy)
5645    if isinstance(value, str):
5646        return Literal.string(value)
5647    if isinstance(value, bool):
5648        return Boolean(this=value)
5649    if value is None or (isinstance(value, float) and math.isnan(value)):
5650        return NULL
5651    if isinstance(value, numbers.Number):
5652        return Literal.number(value)
5653    if isinstance(value, datetime.datetime):
5654        datetime_literal = Literal.string(
5655            (value if value.tzinfo else value.replace(tzinfo=datetime.timezone.utc)).isoformat()
5656        )
5657        return TimeStrToTime(this=datetime_literal)
5658    if isinstance(value, datetime.date):
5659        date_literal = Literal.string(value.strftime("%Y-%m-%d"))
5660        return DateStrToDate(this=date_literal)
5661    if isinstance(value, tuple):
5662        return Tuple(expressions=[convert(v, copy=copy) for v in value])
5663    if isinstance(value, list):
5664        return Array(expressions=[convert(v, copy=copy) for v in value])
5665    if isinstance(value, dict):
5666        return Map(
5667            keys=[convert(k, copy=copy) for k in value],
5668            values=[convert(v, copy=copy) for v in value.values()],
5669        )
5670    raise ValueError(f"Cannot convert {value}")
5671
5672
5673def replace_children(expression: Expression, fun: t.Callable, *args, **kwargs) -> None:
5674    """
5675    Replace children of an expression with the result of a lambda fun(child) -> exp.
5676    """
5677    for k, v in expression.args.items():
5678        is_list_arg = type(v) is list
5679
5680        child_nodes = v if is_list_arg else [v]
5681        new_child_nodes = []
5682
5683        for cn in child_nodes:
5684            if isinstance(cn, Expression):
5685                for child_node in ensure_collection(fun(cn, *args, **kwargs)):
5686                    new_child_nodes.append(child_node)
5687                    child_node.parent = expression
5688                    child_node.arg_key = k
5689            else:
5690                new_child_nodes.append(cn)
5691
5692        expression.args[k] = new_child_nodes if is_list_arg else seq_get(new_child_nodes, 0)
5693
5694
5695def column_table_names(expression: Expression, exclude: str = "") -> t.Set[str]:
5696    """
5697    Return all table names referenced through columns in an expression.
5698
5699    Example:
5700        >>> import sqlglot
5701        >>> sorted(column_table_names(sqlglot.parse_one("a.b AND c.d AND c.e")))
5702        ['a', 'c']
5703
5704    Args:
5705        expression: expression to find table names.
5706        exclude: a table name to exclude
5707
5708    Returns:
5709        A list of unique names.
5710    """
5711    return {
5712        table
5713        for table in (column.table for column in expression.find_all(Column))
5714        if table and table != exclude
5715    }
5716
5717
5718def table_name(table: Table | str, dialect: DialectType = None) -> str:
5719    """Get the full name of a table as a string.
5720
5721    Args:
5722        table: Table expression node or string.
5723        dialect: The dialect to generate the table name for.
5724
5725    Examples:
5726        >>> from sqlglot import exp, parse_one
5727        >>> table_name(parse_one("select * from a.b.c").find(exp.Table))
5728        'a.b.c'
5729
5730    Returns:
5731        The table name.
5732    """
5733
5734    table = maybe_parse(table, into=Table)
5735
5736    if not table:
5737        raise ValueError(f"Cannot parse {table}")
5738
5739    return ".".join(
5740        part.sql(dialect=dialect) if not SAFE_IDENTIFIER_RE.match(part.name) else part.name
5741        for part in table.parts
5742    )
5743
5744
5745def replace_tables(expression: E, mapping: t.Dict[str, str], copy: bool = True) -> E:
5746    """Replace all tables in expression according to the mapping.
5747
5748    Args:
5749        expression: expression node to be transformed and replaced.
5750        mapping: mapping of table names.
5751        copy: whether or not to copy the expression.
5752
5753    Examples:
5754        >>> from sqlglot import exp, parse_one
5755        >>> replace_tables(parse_one("select * from a.b"), {"a.b": "c"}).sql()
5756        'SELECT * FROM c'
5757
5758    Returns:
5759        The mapped expression.
5760    """
5761
5762    def _replace_tables(node: Expression) -> Expression:
5763        if isinstance(node, Table):
5764            new_name = mapping.get(table_name(node))
5765            if new_name:
5766                return to_table(
5767                    new_name,
5768                    **{k: v for k, v in node.args.items() if k not in ("this", "db", "catalog")},
5769                )
5770        return node
5771
5772    return expression.transform(_replace_tables, copy=copy)
5773
5774
5775def replace_placeholders(expression: Expression, *args, **kwargs) -> Expression:
5776    """Replace placeholders in an expression.
5777
5778    Args:
5779        expression: expression node to be transformed and replaced.
5780        args: positional names that will substitute unnamed placeholders in the given order.
5781        kwargs: keyword arguments that will substitute named placeholders.
5782
5783    Examples:
5784        >>> from sqlglot import exp, parse_one
5785        >>> replace_placeholders(
5786        ...     parse_one("select * from :tbl where ? = ?"),
5787        ...     exp.to_identifier("str_col"), "b", tbl=exp.to_identifier("foo")
5788        ... ).sql()
5789        "SELECT * FROM foo WHERE str_col = 'b'"
5790
5791    Returns:
5792        The mapped expression.
5793    """
5794
5795    def _replace_placeholders(node: Expression, args, **kwargs) -> Expression:
5796        if isinstance(node, Placeholder):
5797            if node.name:
5798                new_name = kwargs.get(node.name)
5799                if new_name:
5800                    return convert(new_name)
5801            else:
5802                try:
5803                    return convert(next(args))
5804                except StopIteration:
5805                    pass
5806        return node
5807
5808    return expression.transform(_replace_placeholders, iter(args), **kwargs)
5809
5810
5811def expand(
5812    expression: Expression, sources: t.Dict[str, Subqueryable], copy: bool = True
5813) -> Expression:
5814    """Transforms an expression by expanding all referenced sources into subqueries.
5815
5816    Examples:
5817        >>> from sqlglot import parse_one
5818        >>> expand(parse_one("select * from x AS z"), {"x": parse_one("select * from y")}).sql()
5819        'SELECT * FROM (SELECT * FROM y) AS z /* source: x */'
5820
5821        >>> expand(parse_one("select * from x AS z"), {"x": parse_one("select * from y"), "y": parse_one("select * from z")}).sql()
5822        'SELECT * FROM (SELECT * FROM (SELECT * FROM z) AS y /* source: y */) AS z /* source: x */'
5823
5824    Args:
5825        expression: The expression to expand.
5826        sources: A dictionary of name to Subqueryables.
5827        copy: Whether or not to copy the expression during transformation. Defaults to True.
5828
5829    Returns:
5830        The transformed expression.
5831    """
5832
5833    def _expand(node: Expression):
5834        if isinstance(node, Table):
5835            name = table_name(node)
5836            source = sources.get(name)
5837            if source:
5838                subquery = source.subquery(node.alias or name)
5839                subquery.comments = [f"source: {name}"]
5840                return subquery.transform(_expand, copy=False)
5841        return node
5842
5843    return expression.transform(_expand, copy=copy)
5844
5845
5846def func(name: str, *args, dialect: DialectType = None, **kwargs) -> Func:
5847    """
5848    Returns a Func expression.
5849
5850    Examples:
5851        >>> func("abs", 5).sql()
5852        'ABS(5)'
5853
5854        >>> func("cast", this=5, to=DataType.build("DOUBLE")).sql()
5855        'CAST(5 AS DOUBLE)'
5856
5857    Args:
5858        name: the name of the function to build.
5859        args: the args used to instantiate the function of interest.
5860        dialect: the source dialect.
5861        kwargs: the kwargs used to instantiate the function of interest.
5862
5863    Note:
5864        The arguments `args` and `kwargs` are mutually exclusive.
5865
5866    Returns:
5867        An instance of the function of interest, or an anonymous function, if `name` doesn't
5868        correspond to an existing `sqlglot.expressions.Func` class.
5869    """
5870    if args and kwargs:
5871        raise ValueError("Can't use both args and kwargs to instantiate a function.")
5872
5873    from sqlglot.dialects.dialect import Dialect
5874
5875    converted: t.List[Expression] = [maybe_parse(arg, dialect=dialect) for arg in args]
5876    kwargs = {key: maybe_parse(value, dialect=dialect) for key, value in kwargs.items()}
5877
5878    parser = Dialect.get_or_raise(dialect)().parser()
5879    from_args_list = parser.FUNCTIONS.get(name.upper())
5880
5881    if from_args_list:
5882        function = from_args_list(converted) if converted else from_args_list.__self__(**kwargs)  # type: ignore
5883    else:
5884        kwargs = kwargs or {"expressions": converted}
5885        function = Anonymous(this=name, **kwargs)
5886
5887    for error_message in function.error_messages(converted):
5888        raise ValueError(error_message)
5889
5890    return function
5891
5892
5893def true() -> Boolean:
5894    """
5895    Returns a true Boolean expression.
5896    """
5897    return Boolean(this=True)
5898
5899
5900def false() -> Boolean:
5901    """
5902    Returns a false Boolean expression.
5903    """
5904    return Boolean(this=False)
5905
5906
5907def null() -> Null:
5908    """
5909    Returns a Null expression.
5910    """
5911    return Null()
5912
5913
5914# TODO: deprecate this
5915TRUE = Boolean(this=True)
5916FALSE = Boolean(this=False)
5917NULL = Null()